Writing
Field notes on building AI products end to end.
This is the writing home for my Full-Stack AI Engineer profile: product interfaces, backend workflows, RAG systems, agent control, LLM evaluation, observability, and deployment.
Start here
What Full-Stack AI Engineering Means
A cornerstone essay defining the portfolio’s point of view: AI products need product UX, backend workflow control, data and retrieval systems, model behavior, evaluation, observability, and deployment working together.
Read the essayEditorial promise
- •I write about building AI products as complete systems, not isolated prompts.
- •Good posts should show product context, architecture, trade-offs, failure modes, and implementation direction.
- •The goal is to help engineers and students understand what changes when an AI feature moves from demo to production.
Full-Stack AI Product Notes
How frontend, backend, data, model behavior, evals, and deployment fit together in real AI applications.
RAG That Users Can Trust
Practical notes on ingestion, metadata, retrieval, citations, freshness, and answer review workflows.
Agents With Control
Workflow design for tool use, approvals, traces, state, fallbacks, and operator visibility.
LLM Evaluation in Practice
Golden workflows, rubrics, regression checks, traces, and release gates for AI product teams.
Published writing
Latest notes
AI Radar
Signals worth watching
A small companion stream for AI/ML releases and research signals. This should eventually include my interpretation, not only feed summaries.