Notes at the edge of intelligent systems.
I build enterprise AI — copilots, evaluation frameworks, agentic infrastructure. Before that: production ML at Mastercard and B2B search at G2. I write here when something is worth thinking through in public.
Recent writing
- The AI That Thinks in Physics, Not Paragraphs — World models, why LLMs plateau, and the path forward
- More posts coming soon
What this is
Short and long-form pieces on things I’m actually working through:
- LLM evaluation — what to measure, how to measure it, why most evals are wrong
- Enterprise AI — the grounding problem, retrieval design, agentic systems in production
- Research breakdowns — translating papers into engineering decisions
- Model training — distillation, synthetic data, domain-specific codegen
No content calendar. No growth hacks. Posts appear when there’s something worth saying.
Find me
- LinkedIn — professional updates