KyoMel / Yogyakarta, Indonesia
Daily Termssince June 2025
Definitions, worked examples, and the reasoning between them, from my daily engineering journal. Search by name or by phrase, jump by month, or scan the A-Z index.
Jump17 months
October 202612 terms
Speculative decoding is an inference optimization for autoregressive text generation. A small draft model proposes several tokens. The large target model then checks all proposed tokens in one forw...
Consistent hashing maps each key to a node without a global lookup table. It uses one rule: hash the keys and the nodes onto the same circle, then send each key to the first node clockwise from the...
A vector clock is a logical clock made of a vector of counters. The system has N nodes. Each node keeps one counter per node in the system. A node updates its own counter and copies the counters it...
Data gravity is the pull that a body of data exerts on applications, services, and other data. Dave McCrory coined the term in 2010. He copied the shape of Newton's law. Mass attracts mass, and mas...
PagedAttention is an attention algorithm that keeps the KV cache in fixed-size blocks. The blocks do not sit next to each other in memory. Each sequence owns a block table. The table maps the logic...
The roofline model puts an upper bound on the speed of one kernel on one piece of hardware. It needs only two hardware numbers. The first number is peak compute (Ppeak). This is the top rate of flo...
Rotary Position Embedding (RoPE) is the position scheme that most modern large language models use. It comes from RoFormer (Su et al., 2021, arXiv 2104.09864). RoPE rotates the query vector by the...
A blue-green deployment runs two identical production environments. At any moment, one environment serves all user traffic. The other environment holds the next release. You test the new release in...
Amdahl's Law predicts the speedup of one fixed task when you add workers to it. The law holds one idea: the part of the task that cannot run in parallel sets the ceiling for the whole task.
Dijkstra's algorithm finds the shortest path from one source node to every other node in a weighted graph. Every edge carries a cost. The cost is called a weight, for example minutes, kilometers, o...
Anycast is a network addressing and routing method. One IP address belongs to many servers. The servers sit in different locations. The routing system sends each user to one server. The user does n...
RLHF is a post-training method. It tunes a language model with human preference data. The method does not change the architecture. It changes the weights. A base language model predicts the next to...
September 202622 terms
CQRS is an architectural pattern that separates the operations that change state (Commands) from the operations that read state (Queries), giving each its own model, its own code path, and often it...
Test-Time Compute is the practice of deliberately spending more computation at inference time (when the model is answering) to get a better answer — as opposed to spending more computation at train...
Serverless Cold Start is the latency penalty a serverless platform pays — and your users feel — when it must create and initialize a brand-new execution environment (sandbox, runtime, and your code...
Structured Concurrency is a programming model that guarantees no concurrent task can outlive the code block that created it: whenever a block spawns child tasks, those children must finish — or be...
LoRA is a parameter-efficient fine-tuning (PEFT) technique that adapts a large pre-trained model (an LLM, diffusion model, or any Transformer) to a new task without touching the original weights. I...
An idempotency key is a client-generated unique identifier sent with a mutating HTTP request (typically as an Idempotency-Key header on POST/PATCH) that lets the server recognize retries of the sam...
The Transactional Outbox Pattern makes publishing an event as reliable as committing a database transaction — by not publishing from the application at all. Instead of sending a message to a broker...
Confidential Computing is a hardware-enforced runtime isolation model: a workload runs inside a Trusted Execution Environment (TEE) — a cryptographically sealed region of CPU and GPU memory whose e...
Continuous batching — also called iteration-level scheduling, in-flight batching (TensorRT-LLM) or persistent batching (LMDeploy) — is the LLM-serving scheduler design in which the batch sitting on...
Cell-Based Architecture (also called cellular architecture, cell-based design, or deployment stamps in Azure's vocabulary) is a fault-isolation pattern in which a workload is split into multiple co...
The Circuit Breaker Pattern is a resilience pattern in which every outbound call to one specific dependency is wrapped in a small state machine that watches recent outcomes and, once that dependenc...
Semantic Caching is a cache that matches requests by meaning, not by text. The cache turns every incoming prompt into an embedding vector. It then searches for the nearest prompt it has answered be...
Copy-on-Write (COW) is a memory-sharing rule. Two owners point to the same physical page. The kernel marks that page read-only for both owners. The first write makes a private copy for the writer o...
A CRDT (Conflict-Free Replicated Data Type) is a data type that many replicas update at the same time, without a central coordinator. Any two replicas merge into one identical result. The merge ord...
FlashAttention is an attention algorithm that never materializes the N x N score matrix in GPU global memory. It computes attention in small tiles that stay on-chip, and it fuses the whole operatio...
Backpressure is a control signal that travels backward through a pipeline. A slow consumer sends the signal upstream. A fast producer receives the signal and slows down, blocks, or stops. The consu...
iouring is the asynchronous I/O interface of the Linux kernel. The application writes I/O requests into a ring buffer. The kernel reads them from that shared memory. The kernel writes the results i...
A hedged request is a second copy of a request. The client sends this copy to another backend while the first copy is still in flight. The client uses the first answer that arrives. Then the client...
Constrained Decoding is an inference-time technique that forces a language model to emit only text that is valid against a given grammar, JSON schema, or regular expression. The decoder does this b...
Erasure coding stores an object as fragments, not as whole copies. The encoder cuts the object into k data fragments. Then it computes m parity fragments from the data fragments. It writes the n =...
Model Quantization is the practice of storing a model's numbers in fewer bits than the training format uses. A model trained in BF16 keeps 16 bits for every weight. Quantization stores the same wei...
Write skew is a serialization anomaly. Two concurrent transactions read the same set of rows. Each transaction makes a decision from that read. Then each transaction writes a different row. The wri...
August 202621 terms
Stateless MCP refers to the design and operation of Model Context Protocol (MCP) servers that do not retain any memory or session state between requests — each request is handled independently, in...
A Hybrid Stack is an architecture that combines two or more fundamentally different technologies to solve a problem that no single technology handles optimally on its own. The core idea is "best to...
The Tectonic AI Platform is an architectural approach — and an organizational philosophy — for taming the app sprawl and data fragmentation that result from rapidly adopting AI-assisted development...
The Retry Budget Pattern is a resilience strategy that caps the total amount of retry effort a system is willing to spend on a given operation — before it gives up and fails fast. Instead of "retry...
Tenant-Aware Scaling is the ability of a multi-tenant system to scale its resources based on the specific needs and load of individual tenants, rather than scaling uniformly for all tenants togethe...
TF-IDF is a numerical formula that measures how important a word is to a document within a collection of documents (a corpus). It's one of the most fundamental techniques in information retrieval a...
WebMCP is a proposed web standard (from Chrome and the W3C Web Machine Learning community group) that lets websites expose structured, machine-readable tools for AI agents — so an agent knows exact...
eBPF is a Linux kernel technology that lets you run sandboxed programs inside the kernel without changing kernel source code or loading kernel modules — safely, at native speed, and on live systems...
The ReAct Loop is a prompting and agent-architecture pattern that interleaves Reasoning and Acting in a loop — letting an AI model alternate between thinking through a problem and taking actions (c...
Accelerated Inference is the use of specialized hardware and optimization techniques to run AI model predictions much faster than standard (CPU-only) execution — by offloading the heavy matrix math...
PTAO is a four-stage agent architecture loop that describes how an autonomous AI agent interacts with the world: it Perceives its environment, Thinks about what it knows, Acts to do something, and...
Engineering as a Service (EaaS) is a delivery model where engineering capabilities are provided to an organization as an on-demand service, rather than as a permanent in-house team. Instead of hiri...
The Noisy Neighbor problem is a performance issue in shared, multi-tenant systems where one tenant's heavy or excessive resource consumption degrades the performance of other tenants sharing the sa...
RASP (Runtime Application Self-Protection) is a security technology that is embedded directly into an application and monitors its runtime behavior in real-time to detect and block attacks from ins...
API-Driven Development (also called API-first or API-centric development) is a software development approach where the API is designed first and treated as the primary contract that everything else...
Chain of Responsibility is a behavioral design pattern where a request is passed along a chain of handlers, and each handler decides either to process the request or pass it to the next handler in...
Homomorphic Encryption is a cryptographic technique that lets you perform computations on encrypted data WITHOUT decrypting it first — producing an encrypted result that, when decrypted, matches th...
The Rate-Aware Processing Pattern is a design pattern that respects the rate limits of an external system (an API, a database, a service) while maximizing throughput — by tracking how many requests...
Mixture of Experts (MoE) is a neural network architecture that makes a model dramatically larger in capacity without paying the full cost of running every parameter on every input. It does this by...
A Saga is a sequence of local transactions where each step publishes an event or invokes a call that triggers the next step. If a step fails, the saga runs compensating transactions to undo the alr...
Chaos Engineering is the discipline of running controlled, hypothesis-driven experiments that intentionally inject faults (failures, latency, overload, oracles) into a running production system, in...
July 202622 terms
Loop Engineering is the discipline of designing, optimizing, and debugging the iterative feedback loops that an AI agent uses to reason, act, and learn from results. Instead of a single linear prom...
The modern Senior Developer role has silently split into two jobs disguised as one. The first half is the traditional engineering role we all trained for — systems design, code review, debugging, m...
BASE stands for Basically Available, Soft state, Eventually consistent. It is the consistency model used by distributed systems and NoSQL databases — the practical alternative to ACID (Atomicity, C...
The Codecov Attack was a supply chain breach discovered in April 2021 where an attacker compromised Codecov's Bash Uploader script — a tool used by over 29,000 organizations to upload code coverage...
Artificial Superintelligence (ASI) is a hypothetical future AI that surpasses the best human minds in every domain — not just in narrow tasks like chess or coding, but in general intelligence: scie...
Fully Homomorphic Encryption (FHE) is a revolutionary encryption technique that allows computations to be performed directly on encrypted data without ever decrypting it. The result, when decrypted...
A Columnar Table is a way of storing data on disk where values are grouped by column instead of by row. In a traditional row-oriented database (PostgreSQL, MySQL), every row is stored contiguously...
An Isomorphic Framework (also called Universal JavaScript) is a web framework that lets you write one codebase that runs on both the server AND the client (browser). The same code — same components...
Slopsquatting is a portmanteau of "slop" (AI-generated content, especially low-quality or spammy output) and "cybersquatting" (registering domains that mimic legitimate brands). It describes a grow...
"The Order of Data" is a concept from database systems that describes how the order of query results affects correctness, performance, and user experience — and the hidden dangers of assuming data...
Principal Drift is the gradual separation between a human authority who authorized an action and the machine process that actually carried it out, in systems where AI agents operate autonomously or...
Platform Engineering 2.0 is the evolution of Internal Developer Platforms (IDPs) to serve both humans AND AI agents as first-class citizens — natively supporting AI workloads, embedding FinOps, enf...
Admission Control is a gatekeeper that intercepts requests to a system before they are persisted or executed — and either allows, denies, or mutates them based on predefined policies. It is the las...
"Green Unit Tests" is a critical concept about the false sense of security that comes from seeing all tests passing (green) — when the tests themselves may not actually be validating the right thin...
Attribute-Based Access Control (ABAC) is an authorization model that grants or denies access based on attributes of the user, the resource, the action, and the environment — evaluated against a set...
Engineering Complexity is the study of what makes software hard to build, understand, and change — and the discipline of minimizing it. It distinguishes between two fundamentally different kinds of...
Karate DSL is an open-source test automation framework that combines API testing, UI testing, and performance testing into a single, readable, programming-language-free domain-specific language. In...
Agentic SRE (Site Reliability Engineering) is the evolution of traditional SRE practices to manage systems where AI agents autonomously operate, deploy, and troubleshoot infrastructure alongside hu...
Lift-and-Shift (also called rehosting) is a cloud migration strategy where you move an application from on-premises infrastructure to the cloud with minimal or no code changes. You take the existin...
OpAMP (Open Agent Management Protocol) is a network protocol designed to remotely manage large fleets of telemetry and data collection agents — like OpenTelemetry Collectors, Fluentd, Prometheus ex...
Self-Healing Infrastructure is a system design approach where infrastructure automatically detects, diagnoses, and recovers from failures without human intervention. Instead of waiting for an engin...
Queue-Based Architecture is a design pattern where services communicate asynchronously through an intermediary message queue, instead of calling each other synchronously and waiting for a response....
June 202622 terms
Modulo Hashing is a technique that uses the remainder operator (%) to map any input (like a user ID) into a fixed range of buckets. It's the simplest way to answer: "Given this thing, which slot sh...
LSM Trees are a data structure designed for write-heavy workloads. Instead of updating data in place (like a B-tree), they just append every change to a log, then lazily sort and merge things in th...
U-Net is a convolutional neural network designed for image segmentation — the task of classifying every single pixel in an image. Its signature "U" shape combines a contracting path (which captures...
SSO migration is the process of moving a company's single sign-on system from one identity provider to another — for example, switching from Okta to Microsoft Entra ID — without locking users out,...
The Hi-Lo Algorithm is an ID generation strategy that produces unique identifiers by combining two numbers: a "Hi" value fetched occasionally from a shared source (like a database), and a "Lo" valu...
A Correlation ID is a unique identifier attached to a request that travels with it across every service it touches in a distributed system. It's the breadcrumb trail that lets you trace a single us...
CodingBooth is a tool that gives every project its own isolated, reproducible development environment declared in the repository itself. One command brings it up; one command tears it down. Nothing...
Frame Buffer Hashing is a visual regression testing technique that verifies rendered graphics by reading the raw GPU framebuffer (the pixel data before any compression or encoding), computing a cry...
Perag (short for Personal RAG) is a local, no-cloud, command-line RAG tool that gives your AI assistant long-term memory by making your personal documents searchable. It works entirely on your mach...
A Static Pipeline is a workflow structured as a Directed Acyclic Graph (DAG) where each node is a task and each edge is a dependency (A → B means "A must finish before B starts"). Directed means ev...
WASI-NN is a standardized API specification that lets WebAssembly (WASM) modules perform neural network inference by offloading the computation to native backends — OpenVINO, ONNX Runtime, TensorFl...
HNSW is a graph-based Approximate Nearest Neighbor (ANN) search algorithm that organizes high-dimensional vectors (embeddings) into a hierarchy of navigable small-world graphs. It builds on two ideas:
FinOps (a portmanteau of Finance and DevOps) is an operational framework and cultural practice that brings engineering, finance, and business teams together to manage cloud spending responsibly. It...
Short, lightweight document that captures a significant architectural decision + context + rationale + trade-offs "Document the decision, not the details" Lifecycle: Proposed → Accepted → Deprecate...
TypeScript language extension by Dominic Gannaway (ex-React/Svelte core, creator of Inferno, Lexical, Ripple) "Spiritual successor to JSX" — statement-based instead of expression-based
Distance field: grid where each cell contains distance to nearest obstacle BFS (Breadth-First Search) algorithm to compute this field Used in pathfinding, collision detection, shadow mapping
Human-in-the-Loop (HITL) is a design pattern where a human being is embedded in the decision loop of an automated system — reviewing, validating, correcting, or approving outputs before they take e...
Phantom APIs are undocumented, unapproved, and often unknown API endpoints that exist in production but were never formally reviewed, spec'd, or added to the service catalog. They are "phantom" bec...
Fog Computing is a decentralized computing architecture that extends cloud capabilities to the edge of the network — sitting between the cloud (centralized data centers) and the edge devices (senso...
A Congestion Bug is a class of software defect where a system functions correctly under normal or low load but degrades sharply — or fails entirely — when multiple requests, threads, or processes c...
Tenant Isolation is a security and architecture principle in multi-tenant systems where each customer (tenant) operates in a logically or physically separated environment so that one tenant cannot...
Context Debt is the accumulation of unresolved, fragmented, or outdated context that an AI agent (or developer) must carry forward across interactions — slowing down reasoning, increasing errors, a...
May 202621 terms
The AI Development Lifecycle (ADLC) is the end-to-end, iterative process used to plan, build, train, deploy, and maintain artificial intelligence systems. Because an AI model learns its logic from...
A Smart Datastream is a continuous, real-time flow of data that is analyzed, filtered, enriched, or acted upon while still in motion—rather than simply dumping raw records into a database for later...
AgentOps is the operational discipline and set of practices required to deploy, monitor, evaluate, govern, and continuously manage autonomous AI agents in production. It represents the next evoluti...
The Virtuous Cycle is a self-reinforcing feedback loop in platform engineering where automated reliability, developer ergonomics, and operator ergonomics continuously strengthen one another. Rather...
The Web of Trust is a decentralized model for verifying the authenticity of public keys in PGP (Pretty Good Privacy) and OpenPGP systems. Instead of relying on a single central authority—like a tra...
An A2UI is an emerging interface paradigm designed for collaboration between autonomous AI agents and human users. Unlike traditional UIs where a person must know exactly where to click and what to...
NL2SQL is the process by which an AI system translates a user’s plain-language question into a structured, executable SQL query. It allows non-technical users to interact with relational databases...
CUPID is a set of five software design principles—Composable, Unix-inspired, Predictable, Idiomatic, Domain-driven—created by Daniel Terhorst-North as a practical alternative to traditional OOP gui...
Static Single Assignment (SSA) is an intermediate representation (IR) used by compilers where every variable is assigned exactly once. If a program logically needs to update a value, the compiler i...
A logic gate is the fundamental building block of digital circuits. It takes one or more binary inputs (0 or 1, representing off/on or false/true) and produces a single binary output according to a...
A Zombie API is an endpoint or service that remains exposed and operational long after its intended lifespan—often forgotten, undocumented, and unowned—yet continues to respond to traffic. It is ne...
MTBF — Mean Time Between Failures The average time a system runs correctly before its next failure. It measures reliability. A higher MTBF means the system breaks less often.
A deliberately created set of teamwork rules established when two teams with different management styles merge. Instead of forcing one side to adopt the other’s habits, leaders define a new shared...
3LO is an authorization pattern where a user grants a third-party application limited access to their data hosted by a service provider—without ever sharing their password. The “three legs” are the...
Tactical DDD is the code-level toolbox of Domain-Driven Design. It gives you a set of software patterns to structure the inside of a single Bounded Context so that the code reads like the business...
Apache ActiveMQ is an open-source message broker — think of it as a post office for software applications. It lets different programs (or parts of a program) talk to each other by sending and recei...
Apache Iceberg is an open-source table format for huge datasets. Think of it as a smart filing cabinet for your data lake — it keeps track of all your data files, makes queries fast, and lets you s...
MDASH is Microsoft's new AI-powered vulnerability-finding system that uses 100+ specialized AI agents working together (not just one model) to scan source code, debate whether a bug is real, and ev...
SynthID is Google DeepMind's invisible watermarking tool that embeds a hidden, imperceptible digital signature directly into AI-generated content (images, audio, text, and video) at the moment of c...
Feature flag debt is the accumulation of old, forgotten feature flags (also called "feature toggles") that were used to gradually roll out or test a feature, but were never cleaned up after the fea...
Contract-First Integration is an approach where you design the agreement between services before you write any implementation code. Think of it as drafting a legal contract that both parties sign b...
April 202620 terms
Garbage First Garbage Collector (G1GC) is a region-based, server-oriented garbage collector introduced in Java 7 and made the default GC in Java 9 — designed to provide high throughput with predict...
AI-Assisted Testing is the practice of using artificial intelligence and machine learning models — including large language models (LLMs), code generation models, and intelligent analysis engines —...
Load Shedding Architecture is a deliberate system design strategy where a service or system intentionally drops, rejects, or degrades a portion of incoming requests when it detects it is operating...
GeoDNS (Geographic Domain Name System) is a DNS-based traffic routing technique that returns different IP addresses in response to DNS queries based on the geographic location of the requester — di...
Real-Time Bidding (RTB) is an automated digital advertising process where ad impressions are bought and sold through instant auctions that occur in the time it takes a webpage to load — typically u...
Toy API Calls is a software development learning and prototyping technique where a developer makes the simplest possible, stripped-down API request — using hardcoded values, minimal parameters, and...
Graph Neural Networks (GNNs) are a class of deep learning models designed to operate directly on graph-structured data — learning representations of nodes, edges, and entire graphs by iteratively a...
Cognitive Debt is the accumulated mental burden carried by a person, team, or organization as a result of deferred understanding, unresolved complexity, undocumented decisions, and accumulated ambi...
CoAP (Constrained Application Protocol) is a lightweight web protocol designed for resource-constrained devices like IoT sensors, embedded systems, and low-power networks. Think of it as "HTTP for...
Quorum Algorithms are a class of distributed systems coordination algorithms that ensure consistency and fault tolerance by requiring that any operation — read, write, or decision — be acknowledged...
The Architecture Tax is the ongoing, compounding cost paid by every person and every feature in a software system as a direct consequence of architectural decisions made earlier — where the system'...
The Naked Objects Pattern is a software architectural pattern where the user interface is automatically generated directly from the domain model — with no manually written presentation layer, no vi...
The Fact-Based Labeling (FBL) Framework is a content governance architecture pattern that fundamentally reframes the role of AI classifiers in human review workflows — shifting AI from a judge that...
The SPACE Framework is a multidimensional framework for measuring developer productivity, introduced by researchers from GitHub and Microsoft Research in 2021. It was created to challenge the wides...
Q-Learning is a model-free, off-policy reinforcement learning algorithm that teaches an agent to make optimal decisions in an environment by learning a Q-value function — a table or approximation t...
The Cloud Bandwidth Tax is the compounding, often invisible financial cost imposed on every application, team, and organization that runs workloads in cloud environments — arising from the delibera...
Frontier Agents are the third level in the five-tier hierarchy of AI agent autonomy — autonomous AI systems that go decisively beyond standard task-executing agents by incorporating self-awareness,...
Client ID Metadata Documents (CIMD) is an OAuth 2.0 identity mechanism — currently an IETF Internet-Draft (draft-ietf-oauth-client-id-metadata-document-01, by Aaron Parecki and Emelia Smith, adopte...
An Active Directory (AD) Password is the secret credential assigned to a user account in Microsoft's Active Directory — a centralized system used by organizations to manage who can access computers...
A Billion Laughs Attack is a type of Denial-of-Service (DoS) attack that exploits an XML parser’s entity expansion feature. An attacker submits a very small XML document that defines nested entitie...
March 202622 terms
Graceful Degradation is a design principle where a system continues to function at a reduced level when part of it fails — rather than breaking completely. The goal is to keep core functionality al...
Cosine Similarity is a metric that measures how similar two things are by calculating the cosine of the angle between their vector representations — focusing on direction, not size.
JSON Bandwidth Inflation is the phenomenon where JSON, being a textual format, tends to be verbose, which results in increased network bandwidth usage and higher latencies. In simple terms — JSON w...
Union Find (also known as Disjoint Set Union / DSU) is a data structure that efficiently tracks which elements belong to the same group, and supports two core operations — merging groups and checki...
A Bit Flip is when a single binary digit (bit) unexpectedly changes its value — from 0 to 1, or from 1 to 0 — due to hardware faults, radiation, or other external forces, causing silent data corrup...
Prim's Algorithm is a greedy algorithm used to find the Minimum Spanning Tree (MST) of a weighted, undirected graph. It builds the MST by starting from any node and greedily adding the cheapest edg...
Kruskal's Algorithm is a greedy algorithm that finds the Minimum Spanning Tree (MST) of a graph — connecting all nodes together with the lowest possible total edge weight, using no cycles.
SLI, SLO, and SLA are three related concepts used in Site Reliability Engineering (SRE) to define, measure, and commit to the reliability and performance of a service. Think of it as: SLI is what y...
DPoP (Demonstrating Proof-of-Possession) is an OAuth 2.0 security extension defined in RFC 9449 that cryptographically binds access tokens to a specific client using a public/private key pair — so...
The SECI Model, developed by Ikujiro Nonaka and Hirotaka Takeuchi, is a knowledge management framework that describes how knowledge is created, shared, and transformed within an organization throug...
ASPA (Autonomous System Provider Authorization) is a cryptographic security mechanism for BGP (Border Gateway Protocol) that allows an Autonomous System (AS) to publicly declare which upstream prov...
The Ampere Performance Toolkit (APT) is an open-source performance benchmarking and optimization suite developed by Ampere Computing designed to help developers measure, analyze, and tune applicati...
Comprehension Debt is the growing gap between how much code exists in a system and how much of it any human being genuinely understands — a hidden cost that accumulates when teams rely heavily on A...
IBM MQ Web Server (also called mqweb) is a lightweight, embedded web server component built into IBM MQ that exposes a REST API and a browser-based administrative console (IBM MQ Console) — allowin...
Dynamic Resource Allocation (DRA) is a Kubernetes API and framework introduced in Kubernetes 1.26 (as alpha) that provides a flexible, extensible mechanism for requesting, scheduling, and sharing s...
SOAR (Security Orchestration, Automation, and Response) is a category of security platforms that helps organizations collect threat data, coordinate security tools, automate repetitive tasks, and m...
A Forward Deployed Engineer (FDE) is a highly technical software engineer embedded directly with customers — working on-site or closely alongside a client's team to solve complex, real-world proble...
The DevEx Framework (Developer Experience Framework) is a research-backed model developed by Abi Noda, Margaret-Anne Storey, Nicole Forsgren, and Michaela Greiler — published in ACM Queue and IEEE...
The Facade Pattern is a structural software design pattern that provides a simplified, unified interface to a complex subsystem — hiding the internal complexity of multiple classes, libraries, or A...
Custom Resource Definitions (CRDs) are a Kubernetes extension mechanism that allows you to define your own custom resource types — beyond the built-in Kubernetes objects like Pods, Deployments, and...
WebTransport is a modern web protocol that provides low-latency, bidirectional communication between browsers and servers over HTTP/3 (QUIC). It combines the best features of WebSockets (real-time...
Infrastructure for Agency is a concept coined by Matthew Skelton and Manuel Pais — co-authors of Team Topologies — and articulated by AI Product Advisor Stuart Winter-Tear — that describes the rule...
February 202620 terms
Cloud Systems Drift (also called Configuration Drift or Infrastructure Drift) occurs when the actual state of your cloud infrastructure gradually diverges from its intended or documented configurat...
The Heartbeat is a pattern in distributed systems where nodes (servers, services, or processes) send periodic signals (heartbeats) to indicate they are alive and functioning properly. If a heartbea...
Stackless Continuations are a programming technique that allows a function to pause its execution, save its current state (without saving the entire call stack), and resume later from exactly where...
Container Storage Interface (CSI) is a standardized API that allows container orchestration systems (like Kubernetes) to communicate with any storage system (cloud storage, network storage, local d...
Multi-ERP RAG (Retrieval-Augmented Generation across multiple Enterprise Resource Planning systems) is an AI architecture that allows a Large Language Model (LLM) to query and synthesize informatio...
Semantic Contracts are implicit, unwritten agreements about the meaning and behavior of an API or interface that go beyond the formal technical contract (data types, function signatures). While a t...
TimSort is a hybrid, stable sorting algorithm that combines the best aspects of Merge Sort and Insertion Sort. Designed by Tim Peters in 2002 for Python, it identifies naturally ordered sequences (...
Mean Time to Resolution (MTTR) is the average time it takes to completely resolve an incident or problem from the moment it's detected until it's fully fixed and verified. It measures the total lif...
BEHILOS (BEst HIghest LOwest String) is a challenging benchmark designed to test the reasoning capabilities of Large Language Models (LLMs). It evaluates whether models can correctly identify the s...
Server-to-Server (S2S) Tracking is a method of collecting and transmitting user event data (clicks, conversions, purchases) directly between servers, bypassing the user's browser entirely. Unlike t...
Developer-Centric Cloud Architecture Framework (DCAF) is an approach to designing cloud infrastructure that prioritizes developer experience, productivity, and autonomy. Unlike traditional infrastr...
Sierra Charts is a professional-grade desktop trading and charting platform designed for active traders, particularly in futures, stocks, and forex markets. Unlike web-based platforms, it's a high-...
Exponential Backoff is a retry strategy where the wait time between retry attempts increases exponentially (typically doubling) after each failure, up to a maximum limit. When a request fails (netw...
JSP Code (JavaServer Pages Code) refers to the programming elements embedded within JSP files that enable dynamic content generation on web servers. It combines HTML markup with Java code snippets...
Dutch National Flag is an array sorting pattern (algorithm) that efficiently partitions an array into three sections based on a pivot value: elements less than pivot (left), equal to pivot (middle)...
Multi-Region Active-Active ArchitectureMulti-Region Active-Active Architecture is a distributed system design where your application runs simultaneously in multiple geographic regions (data centers...
Strangler Fig Pattern is a software migration strategy that gradually replaces a legacy system by incrementally building a new system around it, piece by piece, until the old system can be safely r...
Ephemeral Runners are short-lived compute instances (virtual machines or containers) that are created on-demand to execute a single CI/CD job or task, then immediately destroyed after completion. U...
Gaps and Islands Pattern is a SQL technique for identifying consecutive sequences (islands) and missing values (gaps) in ordered data. Islands are continuous ranges of values without breaks (e.g.,...
BFF (Backend For Frontend) is an architectural pattern that introduces a dedicated backend layer for each specific frontend client (web, mobile, etc.), acting as a middleman between the frontend an...
January 202621 terms
Principle of Least Surprise (also called Principle of Least Astonishment) is a design principle stating that a system should behave in a way that users expect, minimizing confusion and unexpected o...
The Sieve Algorithm (specifically Sieve of Eratosthenes) is an ancient, efficient method for finding all prime numbers up to a given limit. It works by systematically "crossing out" multiples of ea...
AutoML (Automated Machine Learning) Frameworks are software tools that automatically handle the complex, time-consuming parts of building machine learning models. They automate tasks like data prep...
Spot Instances are heavily discounted cloud computing resources (virtual machines) that use spare capacity from cloud providers. They can save you up to 90% compared to regular On-Demand pricing, b...
Telemetry Engineering is the practice of designing, implementing, and maintaining systems that automatically collect, transmit, and analyze data from remote sources (applications, servers, devices,...
Bring Your Own Language Model (BYOLM) is a framework or platform feature that allows users to integrate their own custom-trained or third-party language models into a system, rather than being limi...
The Leaky Bucket Algorithm is a rate-limiting technique that controls the flow of data or requests by maintaining a constant output rate, regardless of how bursty the input is. Incoming requests fi...
Dynamic Application Security Testing (DAST) is a security testing method that analyzes a running application from the outside (like a hacker would) to find vulnerabilities. It tests the application...
MongoBleed is a critical security vulnerability in MongoDB that allows attackers to leak sensitive data from server memory through crafted queries. Similar to the infamous Heartbleed bug, it exploi...
The C4 Model is a "maps-of-your-code" approach to software architecture. Created by Simon Brown, it helps teams describe software architecture at different levels of abstraction, much like how you...
SAFe (Scaled Agile Framework) 5.0 is a structured methodology for implementing Agile practices at enterprise scale. It provides a knowledge base of integrated principles, practices, and competencie...
HyperLogLog (HLL) is a probabilistic algorithm used to estimate the cardinality (number of unique elements) in a very large dataset using minimal memory. Instead of storing all unique values, it us...
Passwordless Authentication is a security method that verifies user identity without requiring a traditional password. Instead, it uses alternative factors like biometrics, magic links, one-time co...
High Performance Storage refers to storage systems designed to deliver extremely fast data access, high throughput, and low latency for demanding workloads. These systems optimize for speed, concur...
Holistic Engineering is an approach to software development that considers the entire system and its broader context rather than focusing solely on technical implementation. It emphasizes understan...
The Open-Closed Principle is one of the five SOLID principles of object-oriented design. It states: "Software entities (classes, modules, functions) should be OPEN for extension but CLOSED for modi...
Socket Saturation occurs when a CPU socket (the physical slot on a motherboard where a processor plugs in) reaches its maximum capacity for handling data, computation, or I/O operations. The socket...
Spatial Indexing Architecture is a data structure design that organizes geographic or multidimensional data to enable fast queries like "find all objects near this location" or "what intersects wit...
Data Exfiltration is the unauthorized transfer of data from a computer or network to an external location. It's essentially data theft—stealing sensitive information and sending it somewhere it doe...
ENIAC was the first general-purpose, fully electronic digital computer, completed in 1945 at the University of Pennsylvania. It was built to calculate artillery firing tables for the U.S. Army duri...
Independent Scaling of Components is an architectural principle where different parts of a system can be scaled (increased or decreased in capacity) separately, without affecting other parts. Inste...
December 202523 terms
Dark Deployments (also called Dark Launches) is a deployment strategy where new features or code changes are released to production but remain hidden from regular users. The new functionality runs...
Tromboning is a network inefficiency where traffic takes an unnecessarily long, looping path to reach its destination - going far away only to come back to a nearby location. The name comes from th...
Incremental Data Ingestion is a data loading strategy where only new, modified, or deleted data is captured and transferred since the last ingestion cycle, rather than reloading the entire dataset...
Point of Presence (PoP) is a physical location where network infrastructure is deployed to provide local access points for users. It contains servers, routers, switches, and other equipment that br...
Incremental Static Regeneration (ISR) is a web development strategy that allows static pages to be updated after the site has been built, without requiring a complete rebuild. It combines the speed...
Profile-First System is an architecture approach where user profiles serve as the central foundation for all system functionality, personalization, and decision-making. Instead of building features...
TypedArrays are array-like objects in JavaScript that provide a mechanism for reading and writing raw binary data in memory buffers. Unlike regular arrays that can hold mixed data types, TypedArray...
Cache Stampede (also called Thundering Herd or Dog-pile Effect) is a phenomenon where multiple requests simultaneously attempt to regenerate the same cached data when it expires or becomes unavaila...
Principle of Least Privilege (PoLP) is a security concept where users, applications, and systems are granted only the minimum permissions necessary to perform their specific tasks - nothing more. B...
Commercial ERP (Enterprise Resource Planning) is a pre-built, vendor-developed software suite that integrates and manages all core business processes - including finance, HR, manufacturing, supply...
Secure Software Development Lifecycle (SSDLC) is a framework that integrates security practices into every phase of software development - from planning to deployment and maintenance. Instead of tr...
Jaro-Winkler Distance is a string similarity metric that measures how similar two strings are, with extra weight given to matching prefixes (characters at the beginning). It produces a score betwee...
The Polling Paradox is the counterintuitive problem where traditional polling approaches force you to choose between responsiveness and efficiency, but optimizing for one sacrifices the other. Poll...
Manacher's Algorithm is a clever technique for finding the longest palindromic substring in a string in linear time O(n). It avoids redundant checking by leveraging the symmetry property of palindr...
Vision-Language-Action (VLA) Models are AI systems that directly map visual observations and language instructions to robot actions in a single end-to-end model. Instead of separate systems for see...
TUI is a type of user interface that operates in a text-based environment but uses visual elements like boxes, menus, and buttons drawn with text characters. It's more advanced than a simple comman...
SSOT is a principle where every piece of data or information is stored in exactly one place, and all systems reference that single location. This ensures everyone accesses the same accurate, up-to-...
Design for Partial Failure is a system design principle that assumes some components will fail, so the system is built to continue functioning (possibly with reduced capabilities) instead of comple...
Euclidean Algorithm is an efficient method for finding the Greatest Common Divisor (GCD) of two integers. It works by repeatedly applying the principle that the GCD of two numbers doesn't change wh...
Structured Logging is a practice of writing log messages in a structured, machine-readable format (usually JSON) instead of plain text strings. Instead of embedding data within human-readable sente...
Binary Sovereignty is the principle of maintaining complete control and ownership over the executable code (binaries) that runs on your systems. This means building software from source code rather...
Data Skew occurs when data is distributed unevenly across partitions or nodes in a distributed system, causing some partitions to have significantly more data than others.
First Contentful Paint (FCP) is a web performance metric that measures the time from when a page starts loading to when any part of the page's content is first rendered on the screen.
November 202520 terms
Huffman Coding is a lossless data compression algorithm that assigns variable-length binary codes to characters, with more frequent characters getting shorter codes and less frequent characters get...
JIT Compiler is a compilation technique that compiles code during runtime rather than beforehand. It translates high-level code (like Java bytecode or JavaScript) into native machine code at the mo...
Sandbox Architecture is a security design pattern that creates isolated execution environments where untrusted code can run safely without affecting the host system or other applications. Like a ch...
Conway's Law states that "organizations which design systems... are constrained to produce designs which are copies of the communication structures of these organizations." In simple terms, the sof...
Event-Driven Architecture (EDA) is a design pattern where components communicate through the production and consumption of events. Instead of direct service-to-service calls, systems emit events wh...
PKCE (Proof Key for Code Exchange) is a security extension to OAuth 2.0 designed to make the authorization code flow secure for public clients (like mobile apps and SPAs) that cannot securely store...
Apache Avro is a data serialization framework that provides compact, fast, and language-neutral data exchange. It uses JSON-based schemas to define data structures and serializes data into a compac...
Modular Monoliths are single deployable applications that are internally organized into well-defined, loosely coupled modules with clear boundaries and interfaces. Unlike traditional monoliths wher...
PostgreSQL Vacuum is a maintenance process that reclaims storage space and updates statistics by removing dead tuples (deleted or outdated rows) from tables. Unlike other databases that immediately...
The Floyd-Warshall Algorithm is a dynamic programming algorithm that finds the shortest paths between all pairs of vertices in a weighted graph. It works with both positive and negative edge weight...
Split-Brain Resolution is a mechanism in distributed systems that prevents data corruption and inconsistency when network partitions occur. A "split-brain" happens when cluster nodes lose communica...
The DSPM (Data Security Posture Management) Paradox refers to the counterintuitive situation where organizations implement more data security tools and controls to protect their sensitive data, but...
Persona Scheduler Architecture is a system design pattern that schedules tasks, resources, or services based on different user personas (user types with distinct characteristics, preferences, and b...
Fisher-Yates Shuffle (also called Knuth Shuffle) is an algorithm that randomly shuffles the elements of an array or list, ensuring each possible permutation has an equal probability of occurring. I...
Kadane's Algorithm is a dynamic programming algorithm that finds the maximum sum of any contiguous subarray (consecutive elements) in an array of numbers. It efficiently solves the "Maximum Subarra...
Observability Lasagna is a layered approach to system monitoring and troubleshooting where different types of observability data are stacked like layers in lasagna. Each layer provides a different...
Catastrophic Microservices Migration refers to the disastrous outcome when organizations attempt to break down a monolithic application into microservices without proper planning, strategy, or unde...
Open Policy Agent (OPA) is a general-purpose policy engine that enables unified policy enforcement across different systems and services. It allows you to write policies as code using a declarative...
MITRE ATT&CK Framework is a comprehensive knowledge base that categorizes and describes the tactics, techniques, and procedures (TTPs) that cyber attackers commonly use. It provides a standardized...
Taming Async Chaos refers to strategies and patterns used to manage the complexity, unpredictability, and potential problems that arise when dealing with asynchronous operations in software develop...
October 202521 terms
Delta Encoding is a data compression technique that stores only the differences (deltas) between consecutive values instead of storing the absolute values. This is highly effective when data change...
ID Tokens are JSON Web Tokens (JWTs) used in OpenID Connect (OIDC) to provide identity information about an authenticated user. Unlike access tokens (which grant API access), ID tokens contain clai...
Sticky Session Architecture (also called Session Affinity) is a load balancing technique that routes all requests from a specific user to the same server for the duration of their session. This ens...
A Nondeterministic Finite Automaton (NFA) is a computational model that can be in multiple states simultaneously and can make transitions without consuming input (epsilon transitions). Unlike a DFA...
Behavior Driven Development (BDD) is a software development methodology that focuses on defining and testing software behavior through collaboration between developers, testers, and business stakeh...
Payload Compression is a technique that reduces the size of data being transmitted over a network by encoding it in a more space-efficient format. It decreases bandwidth usage, speeds up data trans...
Speculative Decoding is an optimization technique for large language models (LLMs) that speeds up text generation by using a smaller, faster "draft" model to predict multiple tokens ahead, then hav...
Temporal Debugging Patterns are debugging techniques that help developers understand and troubleshoot issues that occur across time, especially in asynchronous, distributed, or event-driven systems...
Grouped-Query Attention (GQA) is an optimization technique for transformer models that reduces memory usage and computational cost by sharing key-value (KV) pairs across multiple query heads. Inste...
Progressive Scaling Patterns are architectural strategies that allow systems to grow incrementally from simple, single-instance setups to complex, distributed systems without requiring complete rew...
Redlock Algorithm is a distributed locking mechanism designed to create reliable locks across multiple Redis instances. It solves the problem of achieving mutual exclusion in distributed systems by...
Long Short-Term Memory (LSTM) is a type of recurrent neural network (RNN) designed to learn and remember patterns in sequential data over long periods. Unlike traditional RNNs that suffer from vani...
Dead Letter Queue (DLQ) is a service-side queue that stores messages that couldn't be processed successfully after multiple retry attempts. When a message fails repeatedly due to errors, corruption...
Bulkhead Pattern is a fault tolerance design pattern that isolates critical resources and functionality into separate pools to prevent failures in one area from cascading and bringing down the enti...
Two-Phase Commit (2PC) is a distributed transaction protocol that ensures all participants in a distributed system either commit or abort a transaction together, maintaining consistency across mult...
QUIC (Quick UDP Internet Connections) is a modern transport protocol developed by Google that combines the speed of UDP with the reliability of TCP, while providing built-in encryption and improved...
Agent-to-Agent (A2A) Protocol is a communication framework that enables autonomous software agents to interact, negotiate, and collaborate directly with each other without human intervention. These...
Reactive Programming is a programming paradigm focused on handling asynchronous data streams and propagating changes automatically. Instead of imperatively calling functions, you declare how your p...
Stacked Diffs is a version control workflow where you create a series of small, dependent commits that build upon each other in a stack, rather than one large feature branch. Each "diff" (change) i...
Graceful Service Degradation is a resilience pattern where a system continues to operate with reduced functionality when facing failures, rather than completely breaking down. Instead of showing er...
SAML is an XML-based security standard that enables secure single sign-on (SSO) and identity federation between different organizations. It allows users to authenticate once with an Identity Provid...
September 202522 terms
Client-Side Remote Code Execution (RCE) is a security vulnerability that allows an attacker to execute malicious code on a victim's device (browser, mobile app, or client application) by exploiting...
Write Ahead Log (WAL) is a technique used in distributed systems where all changes are first written to a durable log before being applied to the actual data. This ensures data consistency, enables...
Property-Based Testing is a software testing approach that verifies general properties or invariants of your code by automatically generating hundreds or thousands of random test inputs. Instead of...
ECMP (Equal-Cost Multi-Path) is a networking strategy that distributes traffic across multiple paths of equal cost to reach the same destination. In software testing, ECMP testing involves verifyin...
Foreign Data Wrappers (FDW) is a PostgreSQL feature that allows you to access and query data from external sources (other databases, files, APIs, web services) as if they were regular tables in you...
The broker pattern, also known as the intermediary pattern, inserts a middleman called a broker between service users (also known as clients) and service providers (servers). The client is fully di...
AMQP (Advanced Message Queuing Protocol) is an open standard messaging protocol that enables reliable, secure, and interoperable message communication between applications and systems. It provides...
Reactive Scaling is an automatic system behavior where computing resources (servers, containers, etc.) are added or removed in response to current demand or performance metrics. The system "reacts"...
Sociotechnical Design is an approach to software development that considers both technical systems and human/social factors as equally important and interconnected. It recognizes that successful so...
Jakarta Query (formerly JPA Criteria API) is a Java specification that allows developers to build database queries programmatically using Java code instead of writing SQL strings. It provides a typ...
PinConsole is a Unified Internal Developer Platform (IDP) that centralizes all development tools, workflows, and resources into a single, easy-to-use interface. It streamlines the software developm...
Gall's Law states that: "A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched u...
Database-as-a-Service (DBaaS) is a cloud-based service model where a third-party provider hosts, manages, and maintains database infrastructure, allowing users to access and use databases without h...
The Sidecar Pattern is a software architecture design where additional functionality is attached to an application by deploying it in a separate but closely connected container or process, similar...
Recreate Deployment is a deployment strategy where the old version of an application is completely shut down before the new version is deployed. This creates a brief period of downtime during the t...
The Boy Scout Rule in software development states: "Always leave the code cleaner than you found it." Just like Boy Scouts leave campsites better than they found them, developers should improve cod...
he Zero Trust Principle is a cybersecurity philosophy that assumes "never trust, always verify." Unlike traditional security that trusts anything inside the network perimeter, Zero Trust treats eve...
he Read-Replica Pattern is a database architecture where you have one primary (master) database that handles all writes, and multiple read-only copies (replicas) that handle read queries. This sepa...
Serialization converts objects/data structures into a format that can be stored or transmitted (like turning a house into a blueprint). Compression reduces the size of data by removing redundancy (...
Tail Latency refers to the slowest response times in a system - the "long tail" of the latency distribution. While most requests are fast, a small percentage (like the slowest 1% or 5%) can be dram...
A Merkle Tree is a binary tree data structure where each leaf node contains a hash of a data block, and each non-leaf node contains a hash of its two child nodes. This creates a hierarchical struct...
Reservoir Sampling is an algorithm for randomly selecting k items from a stream of n items where n is either very large or unknown. It ensures each item has an equal probability of being selected,...
August 202526 terms
HTTP Pull: Traditional HTTP model where the client initiates requests to the server to retrieve data. The server responds only when requested. HTTP Push: Server-initiated communication where the se...
The Law of Demeter (also known as the Principle of Least Knowledge) is a design guideline that states an object should only communicate with its immediate friends and not with strangers. It promote...
Space complexity is the amount of memory an algorithm needs to run as a function of the size of its input. If an algorithm uses a fixed amount of memory regardless of input size, it has O(1) space...
Time complexity measures how the running time of an algorithm increases as the size of the input grows. If an algorithm examines every element in a list of size n, , its time complexity is O(n).
The 3 Vs are the fundamental characteristics that define Big Data challenges in data engineering: Volume - The amount/size of data Velocity - The speed of data generation and processing
REDCAMEL is an approach for designing APIs and services, consisting of these steps: RE: Requirements — Identify functional and non-functional requirements. DC: Design Considerations — Make architec...
Event-Based Concurrency is a programming model where tasks run concurrently by reacting to events (such as user actions, messages, or I/O), allowing efficient management of multiple tasks without m...
The TLB (Translation Lookaside Buffer) Basic Algorithm handles virtual-to-physical memory address translation by first checking the TLB (a fast, small cache). If the page number is found (TLB hit),...
Linus Torvalds is a software engineer from Finland best known as the creator of the Linux kernel. He was born on December 28, 1969, in Helsinki, Finland. Linus began developing the Linux kernel in...
An Architecture Meta-Frame is a high-level conceptual framework that provides standardized patterns, principles, and guidelines for designing and organizing software architectures. It serves as a "...
Variable shadowing is a technique in which a variable declared within a certain scope has the same name as a variable declared in an outer scope. This is also known as masking. This outer variable...
The Three-Schema Architecture is a database design framework that separates database systems into three distinct levels of abstraction to achieve data independence and system flexibility.
The "No Mapping" Strategy in hexagonal architecture means using the same domain model objects directly across different layers (domain, application, and infrastructure) without creating separate DT...
A sequence or function is monotonically increasing if it never decreases as you move from left to right (or from one term to the next). In other words, each value is greater than or equal to the pr...
A Twitter Snowflake is a unique 64-bit ID system used by Twitter (now X) to generate billions of unique identifiers for tweets, users, and other objects. It's called "Snowflake" because like real s...
The Halting Problem is a fundamental question in computer science: "Given any computer program and its input, can we write another program that determines whether the first program will eventually...
Leaderless Replication is a database strategy where there's no single "master" server controlling writes. Instead, any server (replica) can accept both read and write operations directly from clien...
The PACELC Theorem is an extension of the famous CAP theorem that describes trade-offs in distributed systems. It states: "In case of network Partition, choose between Availability and Consistency;...
Object-Oriented Analysis and Design (OOAD) is a software development approach that models real-world problems using objects (things) and their interactions. It breaks down complex systems into mana...
Adaptive Replacement Cache (ARC) is a smart caching algorithm that automatically balances between keeping recently used items and frequently used items. It "learns" from access patterns and adjusts...
Gossip Protocol is a communication method where nodes in a distributed system spread information by randomly sharing it with a few neighbors at regular intervals, similar to how rumors spread in so...
Memento Design Pattern is a behavioral design pattern that allows you to save and restore an object's previous state without exposing its internal structure. It's like creating "snapshots" or "save...
Bloom Filter is a space-efficient probabilistic data structure that tells you if an element is "definitely not in a set" or "possibly in a set." It can have false positives (says yes when it should...
HSTS (HTTP Strict Transport Security) is a web security mechanism that forces browsers to only communicate with a website over secure HTTPS connections. Once enabled, it prevents users from acciden...
Composition Over Inheritance is a design principle that suggests building complex objects by combining simpler, independent components (composition) rather than creating rigid parent-child class hi...
Server-Sent Events (SSE) is a web standard that allows a server to automatically push real-time updates to a web browser over a single HTTP connection. Unlike WebSockets, SSE is unidirectional (ser...
July 202526 terms
A deadlock is a situation where two or more processes are blocked indefinitely, each waiting for the other to release a resource that they need to continue execution. None of the processes can proc...
Behavioral design patterns are a category of design patterns that focus on the interactions and communication between objects. They help define how objects collaborate and distribute responsibility...
Database Normalization is the process of organizing data in a database to reduce redundancy and improve data integrity by dividing large tables into smaller, related tables and defining relationshi...
Test Driven Development (TDD) is a software development methodology where you write tests before writing the actual code. It follows a simple cycle: Red → Green → Refactor.
Lazy Loading is a design pattern that delays the initialization or loading of resources until they are actually needed, rather than loading everything upfront. This improves performance by reducing...
Black Box Testing is a testing technique that mainly focuses on testing the system functionalities without knowing the actual implementations. There are many testing techniques that actually implem...
Eager Loading is a technique where data or resources are loaded immediately when they are declared or referenced, rather than waiting until they are actually needed. This approach prioritizes upfro...
CARL is a structured interview technique used to evaluate candidates' past experiences and problem-solving abilities through specific storytelling components: Context - The situation or background
RAFT (Replicated And Fault Tolerant) is a consensus algorithm designed to manage distributed systems by ensuring all nodes agree on a shared state, even when some nodes fail. It's simpler and more...
Currying is a technique where a function with multiple arguments is transformed into a sequence of functions, each taking a single argument. Explanation: returns a function that multiplies its argu...
Replica Technique is a design pattern used in distributed systems to enhance reliability, availability, and performance by creating multiple copies (replicas) of data or services across different n...
Functional Programming is a programming paradigm that treats computation as the evaluation of mathematical functions, emphasizing immutability, pure functions, and avoiding side effects.
Event Sourcing is a software architectural pattern where state changes are stored as a sequence of events, rather than just the current state. Each event represents a change in the system, allowing...
Message Persistence is the ability to store messages durably so they survive system failures, restarts, or crashes. It ensures messages are not lost and can be retrieved and processed even after un...
Types of failures in distributed systems refer to the various ways components can malfunction or become unavailable, affecting the system's ability to provide consistent and reliable service across...
"A batch processing system takes a large amount of input data, runs a job to process it, and produces some output data. Designing Data-Intensive Applications, M Kleppmann"
Cloud Exit is the strategic process of migrating applications, data, and IT infrastructure away from cloud services back to on-premises systems or alternative cloud providers to regain control, red...
PCC is a database locking strategy that prevents conflicts by locking data while it's being read or modified, ensuring only one transaction can access it at a time. Use Case
OCC allows multiple transactions to proceed without locks, assuming conflicts are rare; it detects collisions only at commit-time using version tags. Use Case E-commerce carts: many users can edit...
Paxos is a consensus algorithm used in distributed systems to achieve agreement among multiple nodes on a single value, even when some nodes fail or network partitions occur. It guarantees safety (...
Internationalization (i18n) is the process of designing and developing frontend applications to support multiple languages, regions, and cultures without requiring code changes. It enables applicat...
Logical Clocks are mechanisms used in distributed systems to establish a partial ordering of events across different nodes without relying on synchronized physical clocks. They assign timestamps to...
N-Tier Architecture is a software design pattern that separates an application into multiple logical layers (tiers), where each tier has a specific responsibility and communicates only with adjacen...
Big-O notation describes the upper bound of an algorithm's time or space complexity as input size grows. It expresses the worst-case scenario performance, helping developers compare algorithm effic...
Vertical Sharding is a database partitioning technique where columns of a table are split across multiple databases or servers, rather than splitting rows. Each shard contains a subset of columns f...
PIE is a combinatorial counting technique that calculates the size of the union of multiple sets by: Including individual set sizes Excluding pairwise intersections Including triple intersections
June 20253 terms
George Polya developed a systematic approach to problem-solving with four key steps: Understand the problem What is the unknown? What are the given data? \What are the conditions?
gRPC is a high-performance Remote Procedure Call (RPC) framework that can run in various environments and efficiently connect services in and across data centers. By default, gRPC employs Protocol...
A race condition occurs when multiple processes or threads access shared resources simultaneously, and the final outcome depends on the unpredictable timing of their execution.