Jonathan Mckinney - Director Of Research at H2O.ai

Jonathan Mckinney

Director Of Research at H2O.ai

Mountain View, California, United States

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Jonathan Mckinney is a Director of Research at H2O.ai with 17 years of experience building high-performance ML systems, leading development of h2oGPT and GPU-accelerated libraries like H2O4GPU and Driverless AI. Trained as a theoretical astrophysicist (PhD) and former professor and fellow at Stanford and Harvard, he blends deep scientific rigor with production-grade engineering—evident in his significant GPU improvements to the widely used XGBoost project. He specializes in ML system performance, AutoML, multimodal LLM tooling, and scalable GPU parallelism, and he routinely contributes to testing, API integration, and performance-critical backend code. Based in Mountain View, he pairs academic pedigree and prize fellowships with practical open-source impact, steering research into deployable products that emphasize interpretability and private, local LLM deployments.

16 years of coding experience

B.Sc Physics, B.Sc Physics at Texas A&M University

2004 Ph.D; M.S Physics, 2004 Ph.D; M.S Physics at University of Illinois Urbana-Champaign

Russian, English

Github Skills (16)

ccl10

cuda10

xgboost10

machine-learning10

api10

gpgpu10

gpu10

python10

apidoc10

api-design10

gradient-boosting10

accelerated-computing10

cprogramming-language9

distributed-systems9

c-language9

Programming languages (18)

JavaC++CSSRustCGoHTMLSvelte

Github contributions (5)

h2oai/h2ogpt

Apr 2023 - Feb 2025

Private chat with local GPT with document, images, video, etc. 100% private, Apache 2.0. Supports oLLaMa, Mixtral, llama.cpp, and more. Demo: https://gpt.h2o.ai/ https://gpt-docs.h2o.ai/

Role in this project:

Back-end Developer

Contributions:322 reviews, 814 PRs, 6153 pushes in 1 year 10 months

Contributions summary:Jonathan primarily contributed to the development and testing of core features within the h2ogpt repository, with a focus on code testing, summarization, and API integration. They made multiple updates to the test suite and also addressed issues within the API endpoints. The contributions indicate a focus on ensuring the functionality and accuracy of the backend components and improving existing API functionalities.

aichatgptembeddingsgenerativegpt

dmlc/xgboost

May 2017 - Jun 2018

Scalable, Portable and Distributed Gradient Boosting (GBDT, GBRT or GBM) Library, for Python, R, Java, Scala, C++ and more. Runs on single machine, Hadoop, Spark, Dask, Flink and DataFlow

Role in this project:

ML Engineer

Contributions:2 reviews, 9 commits, 14 PRs in 1 year 1 month

Contributions summary:Jonathan significantly improved the performance of the GPU split finding process within the XGBoost library, a core function for gradient boosting. They implemented multi-GPU support using NVIDIA NCCL for the `grow_gpu_hist` histogram method, enhancing the scalability of the library. The user also fixed GPU-related bugs, ensuring accurate functionality, and addressed issues to allow for matrices larger than 2^32 elements. They made further performance improvements by avoiding repeated CUDA API calls and synchronizing only the used GPUs.

xgboostpythonflinkdaskdataflow

Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.

Request Free Trial