Ilan Moscovitz
AI Safety ML Engineer & Researcher at Independent (Advised by William Saunders, Anthropic)
Mountain View, California, United States
Summary
👤
Senior🎓
Top SchoolIlan Moscovitz is an AI Safety ML engineer and researcher with 8 years of experience building production ML systems from YouTube-scale recommendation and search to early-stage startups. He combines hands-on production engineering—distributed metrics pipelines, TFX, and LLM embeddings—with academic-style research on LLM evaluation and deceptive behavior, culminating in an ICLR 2024 paper on black-box lie detection. He authored the interpretable ML library wittgenstein (185k+ downloads), and has built end-to-end evaluation and automated grading pipelines that achieved 93.2% accuracy in detecting problematic model outputs. His background in philosophy (epistemology and symbolic logic) informs rigorous probe design and interpretable methods for both recommender and safety problems. Comfortable translating research into reliable production systems, he focuses on reproducible evaluation, explainability, and tooling that exposes model failure modes before they reach users. Based in Mountain View, he pairs product-facing ML delivery with principled AI-safety research and open-source impact.
8 years of coding experience
2 years of employment as a software developer
Artificial Intelligence, Artificial Intelligence at Georgetown UniversityBachelor of Arts (B.A.), Philosophy (Epistemology and Symbolic Logic), Bachelor of Arts (B.A.), Philosophy (Epistemology and Symbolic Logic) at Grinnell College
Github Skills (2)
fairness-ml10
machine-learning8
Programming languages (1)
Github contributions (5)
imoscovitz/wittgenstein
Nov 2018 - Aug 2021
Ruleset covering algorithms for transparent machine learning
Contributions:114 commits, 2 PRs, 115 pushes in 2 years 8 months
fairness-mltransparentrulesetmachine-learningcovering
imoscovitz/naive_minipy
Feb 2019 - Dec 2021
Contributions:11 pushes, 1 branch in 2 years 10 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.