ML Research: counterfactual cost prediction for blocked tracker requests. Trained production model on 348M requests across 4,592 domains, benchmarked 7 architectures with Tweedie regression and learned URL embeddings. Deploying ONNX model doing live per-request inference in Firefox's anti-tracking engine serving 250M+ users. Also owned Firefox's new tab privacy widget (cross-process C++/JS/IPC service), resolved high-priority security vulnerabilities deep inside Gecko and private browsing, and authored 12+ patches including a feature to SmartBlock and cross-platform CI pipeline across 500M+ installs.
Python, Rust, C++, TypeScript, React, XGBoost, SHAP, conformal prediction, Glean telemetry
