I bridge cutting-edge AI research with production systems. I build things that work — from research prototypes to scalable applications serving real users.
My approach is simple: fast prototyping, measure everything.
- Ship in days, not months - quick iterations reveal the right problems faster than analysis, for both research and products
- Build, measure, learn — whether validating research hypotheses or finding product-market fit
- Eval, Eval and Eval - rigorous benchmarks drive progress. Get the metrics right, the rest follows
- End-to-end AI/ML pipelines from research to production
- Scalable APIs that handle millions of requests
- Real-time AI applications with sub-second latency
- Self-healing infrastructure that scales automatically
- Robust evaluation frameworks and benchmarks
- Systems that gracefully degrade under load
- Documentation that developers like to read
- Products that users actually use
Great research should live in production, not just in papers. I believe the best validation of an idea is when real users depend on it daily.
I focus on:
- Research with immediate utility — Every project should advance knowledge AND ship to users
- Reproducible systems — Code that others can actually run, not just read about
- Scale from day one — Academic prototypes built with production architecture
- Measure what matters — Real-world impact metrics alongside academic benchmarks
I'm always interested in discussing innovative AI applications, teaming up for hackathons, and collaborating on future research.
