Gerson Kroiz is a Senior ML/AI Software Engineer based in Seattle with seven years of experience bridging research and production across ML/NLP, HPC, and full-stack systems. At Google he develops AI code-assist tools and designs resiliency and optimization strategies for large-scale distributed training on TPUs and GPUs, building on earlier work integrating PyTorch Lightning with TPU stacks. His background includes mechanistic interpretability research, hands-on startup CTO experience launching a monetized NLP product, and academic projects in spatiotemporal modeling and scalable numerical methods. Comfortable moving code from prototype to production, he combines deep math and systems intuition—often optimizing parallel workflows and compiler-like performance paths that aren’t obvious from typical ML resumes.
Implementation of the StableLM/Pythia/INCITE language models based on nanoGPT. Supports flash attention, LLaMA-Adapter fine-tuning, pre-training. Apache 2.0-licensed.
Contributions:141 pushes, 16 branches in 11 months
Ongoing research training transformer models at scale
Contributions:590 pushes, 14 branches, 1 comment in 1 year 2 months
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
Gerson Kroiz - Senior ML AI Software Engineer at Google