AI

Writing on AI engineering

Notes and how-tos on building with LLMs — prompting, agents, retrieval, and evaluation. Written to help other engineers ship AI features that hold up in production.

Portrait of Sachin Gupta rendered in binary

Articles

33 articles

Writing· MCP series
Trust and Authorization in MCP
OAuth 2.1, the resource-server model, RFC 8707, and the threats the protocol cannot solve for you.
The security axis of MCP: how authorization works on the HTTP transport with OAuth 2.1, the server-as-resource-server model and RFC 8707 resource indicators, the later OIDC discovery and incremental-consent refinements, and the three threat classes, token passthrough, confused deputy, and prompt injection via tool output, with the guard for each.
- MCP
By Sachin Gupta20 Jun 20268 min read
Writing· Essay
Most Agent Failures Are Loop Failures
The control loop is where you design termination, budgets, and recovery.
When an agent misbehaves, the loop is usually the culprit, not the model. Loop engineering is the deliberate design of when to keep going, when to stop, and what to do when a step fails.
- Loop Engineering
By Sachin Gupta18 Jun 20265 min read
Writing· MCP series
MCP on the Wire
JSON-RPC, the lifecycle handshake, and the two transports, stdio and Streamable HTTP.
How MCP messages actually move. The JSON-RPC base, the initialize handshake and capability negotiation that set up every connection, and the two standard transports, stdio for local servers and Streamable HTTP for remote ones, plus why the old HTTP-plus-SSE transport was replaced.
- MCP
By Sachin Gupta15 Jun 20267 min read
Writing· Search series
Rerankers
Retrieve fast and roughly, then read the top few closely and reorder. The second pass that fixes the first.
The closing part of the search series. First-stage retrieval is fast but ranks crudely. A reranker takes the top candidates and scores each one by reading the query and document together with a cross-encoder, then reorders. This covers the bi-encoder vs cross-encoder split, why it must be a second stage, and where it lives in production, with an interactive retrieve-then-rerank demo.
- Search
By Sachin Gupta14 Jun 20267 min read
Writing· Search series
Hybrid Search
Keyword search and vector search fail in opposite directions. Run both, merge the results, and you keep the strengths of each.
Part six of the search series. BM25 is strong on exact terms and blind to synonyms; vector search is the reverse. Hybrid search runs both and fuses the two ranked lists. The hard part is combining incompatible scores, which Reciprocal Rank Fusion (RRF) sidesteps by using ranks instead. Worked examples plus an interactive RRF explorer.
- Search
By Sachin Gupta12 Jun 20267 min read
Writing· Essay
You Cannot Improve What You Cannot Measure
Operating an LLM system means making a fuzzy, non-deterministic thing measurable.
LLM systems are hard to operate because outputs are open-ended and non-deterministic. LLM ops and eval is the discipline that makes them measurable: evals as tests, tracing, regression gates, and a flywheel that turns production failures into your next test set.
- LLM Ops & Eval
By Sachin Gupta11 Jun 20265 min read
Writing· MCP series
The Six Primitives of MCP
Tools, resources, prompts, sampling, roots, elicitation, organized by who is in control.
The heart of MCP is six primitives: three the server offers (tools, resources, prompts) and three the client offers (sampling, roots, elicitation). This piece defines each precisely, with method names and the one question that organizes them, plus what the 2025-11-25 revision added around them: async tasks, tool-calling in sampling, richer elicitation, and icons.
- MCP
By Sachin Gupta10 Jun 20268 min read
Writing· Search series
ANN at Scale: HNSW
Give up on exact, and you can search a billion vectors in milliseconds. This is the graph that makes it happen.
Part five of the search series. Exact kNN is O(N) and hopeless at scale, so production vector search gives up exactness for speed with approximate nearest neighbor. This explains the dominant method, HNSW: navigable small-world graphs, greedy search, the layer hierarchy, and the M / efConstruction / ef knobs, with an interactive graph-search explorer.
- Search
By Sachin Gupta10 Jun 20268 min read
Writing· Search series
Vector Search (kNN)
When you stop comparing characters and start comparing coordinates, search can finally match meaning.
Part four of the search series. Keyword and fuzzy search match characters, so they miss synonyms. Vector search embeds text into geometry, where similar meaning becomes nearby points. This covers embeddings, the two distance metrics (cosine vs euclidean), exact k-nearest-neighbor search, and the scaling wall that forces the approximate methods in the next part. With an interactive kNN explorer.
- Search
By Sachin Gupta8 Jun 20267 min read
Writing· MCP series
A Field Guide to MCP
What the Model Context Protocol is, the pieces it is made of, and the map of the rest.
The Model Context Protocol is how an AI application reaches the tools and data it needs. By mid-2026 it is a Linux Foundation standard with an official registry and support across the major platforms. This opening piece lays out the problem, the client-host-server architecture, the six primitives, and the map, anchoring a five-part series.
- MCP
By Sachin Gupta5 Jun 20267 min read

Explore by tag

AI — Sachin Gupta