Working on
- OneNote Copilot eval framework at Microsoft — building the LLM-as-a-Judge layer that scores Copilot video responses across truthfulness, helpfulness, and safety.
- A long-form study companion (Engineering Clarity — Data, Distributed Systems and AI) — published as a work-in-progress essay, updated as I learn.
- An early-stage AI-agents / robotics startup. Still pre-product. No URL yet.
Reading
- Designing Data-Intensive Applications — Kleppmann (re-reading the consistency chapters).
- The Pragmatic Engineer newsletter — Gergely Orosz.
- Anthropic and OpenAI research blogs for evaluation and agentic patterns.
Building
- This blog — minimalist v8 rebuild, May 2026.
- A growing notes repo of cloud + ML decision trees that I want to publish here.