Staff Machine Learning Engineer at Coinbase working on applied AI systems and automation. Previously, Senior Applied Scientist at Amazon, where I built production NLP, post-training, and evaluation systems.
I also independently build native software for Apple platforms.
I work across problem definition, experimentation, evaluation, system architecture, and production deployment. My recent work has focused on agentic systems, multilingual NLP, machine translation, model quality, and automated evaluation.
Based in Seattle. Outside work, I enjoy skiing, snowboarding and running.
Enabling the automation of manual compliance investigations with agentic systems. My work spans stakeholder discovery, architecture selection, implementation, and evaluation with the rest of the engineering team.
Built an automated, multi-task post-training workflow for multilingual NLP that combined data preparation, source-mix and dataset-size optimization, Bayesian hyperparameter optimization, standardized evaluation, and human-gated deployment. It reduced model delivery time from months to days and enabled the organization’s custom models to reach production.
Built an automated, layered evaluation system spanning quantitative model comparison, categorical issue classification, and descriptive retrieval-augmented LLM-as-judge analysis. It reduced evaluation time from weeks to minutes.
Built and deployed domain-adapted seq2seq and Transformer models that improved translation quality over generic baselines and served production traffic for years.
A privacy-first macOS app that turns Messages history into local analytics while keeping the analysis on the user’s Mac.
Staff Machine Learning Engineer
Agentic systems and compliance automation.
Senior Applied Scientist, NLP
Production NLP, multilingual machine translation, post-training, and evaluation.
Ph.D., Mathematical Modeling / Numerical Simulation
MBA, Finance & Management
M.S., Engineering
B.S., Engineering
