Home
Zhiyuan Jerry Lin
zylin@cs.stanford.edu · ItsMrLin.com
zylin@cs.stanford.edu · ItsMrLin.com
zylin@cs.stanford.edu
ItsMrLin.com
Principled, data-efficient learning and optimization: developing methods that use scarce human judgment, language, preferences, and behavioral signals to guide adaptive data collection, model evaluation, and AI system improvement.
Education
Stanford University Ph.D. in Computer Science September 2016 - June 2021
Computational social science, human-AI decision making, data science, applied statistics and machine learning.
Georgia Institute of Technology B.S. in Computer Science, Minor in Math; Highest Honor August 2012 - May 2016
Focused on intelligence, info-networks, and theory for CS major; probability and statistics for math minor.
Experience
Research Scientist Meta June 2019 - Present
2026-Present: Applied AI. Drive auto-research and expert-supervised AI improvement initiatives for models and agents, combining active data generation, expert-supervised improvement loops, and targeted evaluation to move beyond low-quality synthetic data and scale scarce expert attention into high-quality data and measurable quality gains.
2021-2026: Adaptive Experimentation, Central Applied Science. Led a major technical direction modernizing Meta's adaptive experimentation and Bayesian optimization systems, developing principled AI-augmented methods that extend black-box optimization beyond fixed numeric/vector spaces to language-rich system components, natural-language goals, human feedback, multiple objectives, and scarce online experiments; applied across large-scale AI and recommender systems, infrastructure optimization, and user-facing products; core contributor to open-source projects BoTorch and Ax.
2019-2020: Part-time Researcher, Core Data Science. Preference learning, multi-objective decision modeling, and Bayesian optimization.
Social Network Research Intern Tencent June 2017 - September 2017
At the WeChat team, built ML models for user behavior and article-forwarding prediction leveraging latent overlapping group structure; improved baseline performance by 20%.
Applied Research Intern eBay 2015 - 2016
Developed ML-based marketplace recommendation models for related and complementary products; improved internal related-product recommendation baseline's performance by 30%.
Media Application & Presentation Engineer Intern Yahoo May 2014 - August 2014
Improved Yahoo Homepage mobile article-end experience, increasing engagement time for the targeted user group by 3x.
Selected Publications
Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts Working paper
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Samuel Dooley,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Samuel Dooley,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Samuel Dooley,
  • Maximilian Balandat,
  • Eytan Bakshy
LILO: Bayesian Optimization with Natural Language Feedback ICML 2026
  • Katarzyna Kobalczyk,
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Zhuokai Zhao,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Katarzyna Kobalczyk,
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Zhuokai Zhao,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Katarzyna Kobalczyk,
  • Zhiyuan Jerry Lin,
  • Benjamin Letham,
  • Zhuokai Zhao,
  • Maximilian Balandat,
  • Eytan Bakshy
Ax: A Platform for Adaptive Experimentation AutoML 2025
  • Meta Adaptive Experimentation team
  • Meta Adaptive Experimentation team
  • Meta Adaptive Experimentation team
Joint Composite Latent Space Bayesian Optimization ICML 2024
  • Natalie Maus,
  • Zhiyuan Jerry Lin,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Natalie Maus,
  • Zhiyuan Jerry Lin,
  • Maximilian Balandat,
  • Eytan Bakshy
  • Natalie Maus,
  • Zhiyuan Jerry Lin,
  • Maximilian Balandat,
  • Eytan Bakshy
qEUBO: A Decision-Theoretic Acquisition Function for Preferential Bayesian Optimization AISTATS 2023
  • Raul Astudillo,
  • Zhiyuan Jerry Lin,
  • Eytan Bakshy,
  • Peter I. Frazier
  • Raul Astudillo,
  • Zhiyuan Jerry Lin,
  • Eytan Bakshy,
  • Peter I. Frazier
  • Raul Astudillo,
  • Zhiyuan Jerry Lin,
  • Eytan Bakshy,
  • Peter I. Frazier
Preference Exploration for Efficient Bayesian Optimization with Multiple Outcomes AISTATS 2022
  • Zhiyuan Jerry Lin,
  • Raul Astudillo,
  • Peter I. Frazier,
  • Eytan Bakshy
  • Zhiyuan Jerry Lin,
  • Raul Astudillo,
  • Peter I. Frazier,
  • Eytan Bakshy
  • Zhiyuan Jerry Lin,
  • Raul Astudillo,
  • Peter I. Frazier,
  • Eytan Bakshy
Probability Paths and the Structure of Predictions over Time NeurIPS 2021
  • Zhiyuan Jerry Lin,
  • Hao Sheng,
  • Sharad Goel
  • Zhiyuan Jerry Lin,
  • Hao Sheng,
  • Sharad Goel
  • Zhiyuan Jerry Lin,
  • Hao Sheng,
  • Sharad Goel
Blind Justice: Algorithmically Masking Race in Charging Decisions AIES 2021
  • Alex Chohlas-Wood,
  • Joe Nudell,
  • Keniel Yao,
  • Zhiyuan (Jerry) Lin,
  • Julian Nyarko,
  • Sharad Goel
  • Alex Chohlas-Wood,
  • Joe Nudell,
  • Keniel Yao,
  • Zhiyuan (Jerry) Lin,
  • Julian Nyarko,
  • Sharad Goel
  • Alex Chohlas-Wood,
  • Joe Nudell,
  • Keniel Yao,
  • Zhiyuan (Jerry) Lin,
  • Julian Nyarko,
  • Sharad Goel
The Limits of Human Predictions of Recidivism Science Advances, 2020
  • Zhiyuan (Jerry) Lin,
  • Jongbin Jung,
  • Sharad Goel,
  • Jennifer Skeem
  • Zhiyuan (Jerry) Lin,
  • Jongbin Jung,
  • Sharad Goel,
  • Jennifer Skeem
  • Zhiyuan (Jerry) Lin,
  • Jongbin Jung,
  • Sharad Goel,
  • Jennifer Skeem