Research with an engineer’s sense of responsibility.

I study machine learning problems where performance, robustness, and reproducibility need to hold at the same time.

How I work

I prefer explicit assumptions, controlled experiments, and results that can be inspected. Engineering is part of the research process: data lineage, repeatable builds, and safe deployment all affect the quality of the evidence.

What this site contains

This homepage is a concise map of active work. Longer research notes, implementation details, and technical experiments live on the blog.

Working principles

• Make uncertainty visible.
• Prefer reproducible evidence over fragile demos.
• Keep systems understandable enough to operate safely.