Chunking strategies that survive production
Structural, recursive, and semantic splitting compared — plus the sizes to start from and the five ways chunking quietly caps your retrieval quality.
Intermediate8 min
#rag
#retrieval
#pipeline
Newly added · AI engineering
A running stream of in-depth pieces on LLM internals, prompting, retrieval, agents, MCP, context engineering, evals, and production observability — each one filed under a category you can browse.
The latest pieces across every category, newest first.
Structural, recursive, and semantic splitting compared — plus the sizes to start from and the five ways chunking quietly caps your retrieval quality.
Listed inWhat are AI Agents?AI Agentson
Anthropic on the difference between workflows and agents, five composable patterns, and why most teams should start with the simplest thing that works.
Listed inTokensHow LLMs Workon
Why character and word counts are wrong by enough to matter, how BPE actually splits your text, and how to budget a request so the model never gets cut off mid-sentence.
Prepending model-generated context to each chunk before embedding, and the measured drop in failed retrievals it produced.
Listed inData LayerModel Context Protocolon
The normative source: lifecycle, capability negotiation, transports, and the exact shape of tools, resources, and prompts.
Listed inModel Context Protocol (MCP)Model Context Protocolon
Hosts, clients, servers, and the three primitives a server can expose — with a working TypeScript server in forty lines and the four mistakes that make one useless.
Listed inModel Context Protocol (MCP)Model Context Protocolon
The announcement that started the ecosystem — why a shared protocol beats N×M bespoke integrations.
Listed inWhat is an AI Engineer?Introductionon
The role is context plumbing, schema design, evaluation, and cost control — not gradients. What you need to know to start, and what you can safely defer.
Listed inMulti-Agent SystemsAI Agentson
Lilian Weng's survey of planning, memory, and tool use — still the clearest map of the agent design space.
Listed inWhat is an AI Engineer?Introductionon
The engineer who builds products on top of existing models rather than training new ones from scratch.
Listed inGoogle GeminiModels & Providerson
Gemini's multimodal-native design and the Vertex AI path to production.
Listed inLong Context ProcessingContext Engineeringon
The paper behind the U-shaped recall curve: models attend to the start and end of a long context and lose the middle.
Listed inPerforming Similarity SearchVector Databaseson
Cosine vs dot product vs L2, top-k, and metadata pre-filtering.
Listed inPrompt Injection AttacksSafety & Ethicson
Years of worked examples showing why filtering does not solve injection, and why the dual-LLM pattern is the closest thing to a fix.
Listed inTransport LayerModel Context Protocolon
stdio and streamable HTTP, and which one your deployment needs.
Pinecone walks through fixed, recursive, and document-aware splitting with concrete size and overlap recommendations.
Listed inDeterministic EvalsEvaluationon
Exact match, schema validation, and assertions that never disagree with themselves.
Listed inHow LLMs WorkHow LLMs Workon
The transformer paper. Worth reading once even if you never train anything — everything downstream inherits its shape.
Listed inInferenceIntroductionon
Running a trained model to get output, and the cost, latency, and throughput levers around it.
Listed inFunction CallingAPIs & SDKson
OpenAI's reference on declaring tools, handling the model's call, strict schemas, and parallel tool use.