I'm passionate about distributed computing, LLM/Gen AI systems and open source. I built multiple 0 to 1 systems and products working with diverse teams, tools and constraints. Alawys optimizing for execution, performance and comprehensive solutions to challenging problems.
Currently, I am a Member of Technical Staff @ OpenAI. Unfortunately, that means most of my current work will be private.
Aside from my own projects, I have contributed to or took a leading role in several open source projects. The most notable include:
- vLLM, A high-throughput and memory-efficient inference and serving engine for LLMs (commiter)
- Ray, an open source framework for building distributed applications (commiter)
- RayLLM, library for LLMs on Ray (maintainer)
- Equestria at War, a Hearts of Iron IV modification (maintainer)
Former/inactive projects:
- PyCaret, an open source, low-code machine learning library in Python (maintainer)
- tune-sklearn, a scikit-learn API on Ray Tune (maintainer)
- xgboost-ray, distributed XGBoost on Ray (maintainer)
- lightgbm-ray, distributed LightGBM on Ray (maintainer)
- hpbandster-sklearn, a scikit-learn API wrapper for HpBandSter (author)
- cwtools, a library for parsing, editing, and validating Paradox Interactive script files (commiter)
- LinkedIn: https://www.linkedin.com/in/yard1/
- Steam Workshop: https://steamcommunity.com/profiles/76561198045710503/myworkshopfiles/