About
Engineering systems, examined in the open
I work across industry-grade machine learning systems and academic investigation, with a steady focus on model optimization, evaluation methodology, and robust ML systems.
Background
A path between systems and research
My work has moved between building production machine learning systems and investigating the questions those systems raise. Industry taught me how models behave under real constraints, latency budgets, and messy data. Academic study pushed me to ask why a method works, where its assumptions break, and how a claim should be measured before it is trusted. That tension between building and testing is what I return to. Each project and paper has sharpened how I approach optimization, evaluation design, and the engineering of reliable systems.
The thread running through it is simple: make the reasoning visible, so the next person can check it, reuse it, or improve it.
Approach
Transparent assumptions. Rigorous evaluation. Honest results.
I try to make every assumption explicit before I make a claim. Benchmarks should be chosen before the results are known, reported in full, and read with their limits in view. When collaborating, I aim for open technical communication, so that a method can be judged on what it actually does.