Summary
Katrina Drozdov is an AI research scientist with a decade of experience building robust, production-minded machine learning systems and evaluations. With a PhD from NYU Center for Data Science and deep roots in Yann LeCun’s group, she has advanced self-supervised learning for computer vision and translated research into practical gains like one-shot learning improvements during Meta internships. Now based in the San Francisco Bay Area, she focuses on benchmarking and rigorous evaluation of large language models to bridge research insights and real-world deployment. She has contributed targeted reasoning tasks at Snorkel AI and currently develops evaluation methodologies at Vals AI to stress-test LLM capabilities. Comfortable moving models from lab to product, she has a track record across industry (eBay, Meta) and academia of delivering scalable, measurable impact. An often overlooked strength is her mathematical foundation from Harvard, which underpins her ability to design principled evaluation metrics that reveal subtle failure modes.
10 years of coding experience
10 years of employment as a software developer
Doctor of Philosophy - PhD, Data Science, Doctor of Philosophy - PhD, Data Science at New York University
Bachelor's degree, Mathematics, Bachelor's degree, Mathematics at Harvard University
Master’s Degree, Data Science, Master’s Degree, Data Science at NYU Center for Data Science
Sofia High School of Mathematics