Topics
Five pillars of building AI systems that hold up in production — and the writing under each.
Evaluation
Metrics whose assumptions match the problem, splits that don’t flatter the model, and trajectory-level evaluation for agents — numbers a person can actually act on.
Quality
Quality on two fronts — model quality (does it hold up, where does it break) and data quality (lineage, sensitivity, and governance over enterprise data estates).
Context engineering
Getting the right information in front of the model at the right moment — retrieval, grounding, and the context that decides whether an answer is any good.
Knowledge base
Knowledge graphs and RAG over enterprise data — turning scattered estates into a queryable, trustworthy source of truth.
Integration
Wiring AI into the systems around it — tools, APIs, MCP, and data pipelines — so a model becomes something a business can actually use.
Open-source & open-weight AI
Building on open models and open-source tooling — for control over cost, privacy, and how a system actually behaves.