— the engineering journal
All writing. 4 essays.
- 01
Building Agents Is Easy. Reliability Is the Real Work.
LatestAn AI agent demo can come together in an afternoon. Making it behave under noisy inputs, ambiguous prompts, and real user behavior is the real discipline. Notes from the unglamorous half of AI engineering — evals, metrics, failure modes, and the build–evaluate–improve loop.
May 202612 min readIssue 08AIEngineeringAI EngineeringEvals - 02
From Chains to Graphs: Why LangGraph Is How I Build Agents Now
Linear chains break the moment your agent needs to retry, branch, or wait for a human. Why I migrated from LangChain chains to LangGraph for every multi-step AI workflow — and what production taught me on the way.
Apr 202611 min readIssue 07AIBackendAI / GenAILangGraph - 03
Building an MCP Server Inside VS Code: JSON-RPC at the IDE Layer
A deep dive into bridging AI agents with the editor using the Model Context Protocol, JSON-RPC 2.0 streaming, and a chat surface that finally feels native to the IDE.
Apr 202612 min readIssue 07AIFrontendArchitectureAI Systems - 04
RAG in Production: What the Tutorials Don't Tell You
Chunking strategies, retrieval quality metrics, eval loops, and the unglamorous infra work that decides whether your RAG actually ships — beyond the demo.
Mar 202614 min readIssue 06AIBackendAI / LLMRAG
that's all of it — for now.
Suggest a topic