I've been building ML systems for over a decade now — shipping models at DARPA, Oracle, Clarabridge, and Qualtrics — but I've rarely taken the time to write about what I've learned along the way.
This blog is an attempt to change that. Expect posts about:
- NLP and LLMs in production — the gap between research and shipping
- Evaluation — how to know if your system actually works
- Tools and workflows — things that have made my work faster or better
- Research I find interesting — papers, ideas, and open questions
Nothing polished. Just thinking out loud.