Sharon Zhou is Vice President of Artificial Intelligence at AMD and the former CEO and co‑founder of Lamini, a generative AI startup whose team and tech transitioned into AMD after raising over $25M at a $136M valuation. She teaches generative AI to nearly a million learners online and at Stanford, builds courses, and regularly speaks to audiences from top researchers to Fortune 500 leaders. A Stanford CS PhD advised by Andrew Ng, she led a 50+ student research group, published award‑winning work (top 1% NeurIPS), and contributed calibration and Brier‑score improvements to the influential BIG-bench benchmark. Equally comfortable in product, research, and startup execution, she focuses on algorithms and agentic data methods that reduce LLM hallucinations in production. Trained in Classics at Harvard (summa cum laude) and fascinated by ideas that last millennia, she blends technical rigor with a storyteller’s instinct — as likely to tweak benchmarking math as to quote Plato.
Beyond the Imitation Game collaborative benchmark for measuring and extrapolating the capabilities of language models
Role in this project:
Data Scientist
Contributions:3 reviews, 13 commits, 2 PRs in 1 month
Contributions summary:Sharon focused on improving metrics and calibration within the `big-bench` project, a benchmark for language models. Their contributions primarily involved modifications to the `task_metrics.py` file, implementing and refining Brier score calculations for evaluating model performance. The user added tests, fixed bugs, and improved the documentation related to the scoring functions. They also ordered and normalized target scores to improve calibration metrics accuracy.
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.