Engineering

LIN F. YANG | 杨林

Dr. Lin Yang is currently an Associate Professor (tenured) in the Electrical and Computer Engineering Department and Computer Science Department at the University of California, Los Angeles.

His research focuses on the foundations of modern machine learning and data science. He develops fast algorithms with provable guarantees, especially in reinforcement learning, large language model acceleration, compression, and quantization, non-convex optimization, and streaming algorithms. He is particularly interested in safe and efficient decision-making under uncertainty, bridging theoretical insights with real-world impact in areas such as healthcare, robotics, and scientific discovery.

Dr. Yang is a recipient of an Amazon Faculty Award, the Simons Research Fellowship, Dean Robert H. Roy Fellowship, and the JHU MINDS Best Dissertation Award. Prior to joining UCLA, he was a postdoctoral researcher at Princeton University working with Prof. Mengdi Wang. He received dual Ph.D. degrees in Computer Science and in Physics & Astronomy from Johns Hopkins University, advised by Alex Szalay and Vova Braverman.

中文简介

[Last updated Jun. 2026]

Research

I study the foundations and systems principles behind reliable learning and decision-making. My recent interests connect reinforcement learning, agentic AI frameworks, LLM fine-tuning, and LLM compression and quantization.

Research map connecting reinforcement learning, agentic frameworks, LLM fine-tuning, and LLM compression A diagram showing four research areas feeding into efficient and reliable AI: reinforcement learning, agentic frameworks, LLM fine-tuning, and LLM compression and quantization. Reinforcement Learning exploration, safety, efficiency Agentic AI Frameworks planning, memory, feedback LLM Fine-Tuning adaptation, alignment, evaluation LLM Compression & Quantization small models, fast inference Efficient, Reliable AI Systems

Reinforcement learningprovable sample efficiency, safe exploration, and robust decision-making under uncertainty

Agentic frameworksAI agents that plan, use tools, remember context, and learn from feedback

LLM fine-tuningtask adaptation while preserving reliability, efficiency, and evaluation rigor

LLM compression and quantizationlower memory and compute costs for faster, broader deployment

Featured work: Winning Gold at IMO 2025 with a Model-Agnostic Verification-and-Refinement Pipeline, with Yichen Huang. We study how verification and iterative refinement can turn strong general-purpose LLMs into reliable mathematical reasoning agents.

Application Domains

Healthcare
Robotics
Science Discovery
Finance

Publication records: Google Scholar / DBLP

Service

Senior Area ChairNeurIPS 2026

Area ChairICLR, AISTATS, ICML, NeurIPS

Senior Program CommitteeAAAI

Conference ReviewingAAAI, ICML, NeurIPS, STOC, FOCS, COLT, PODS, ITCS, ACML, AISTATS, ICALP, RANDOM, ESA, LATIN, KAIS, FSTTCS

Journal ReviewingTheoretical Computer Science, QIC, JCSS, OPRE, TMLR

People

Students

  • Yunfan Li
  • Chang Liu
  • Lawrence Liu

Ph.D. Alumni

Other former advisees
  • Qiwen Cui former undergraduate advisee; Ph.D. from UW
  • Kunhe Yang former undergraduate advisee; Ph.D. student at Berkeley
  • Junhong Shen former undergraduate advisee; Ph.D. student at CMU
  • Jinghan Wang former undergraduate advisee; Ph.D. student at UMD
  • Dingwen Kong former undergraduate advisee; Ph.D. student at MIT

Education

Johns Hopkins University Baltimore, MD

Ph.D. in Computer Science, Advisors: Vladimir Braverman, Alex Szalay; Ph.D. in Physics & Astronomy, Advisor: Alex Szalay; MS.E. in Computer Science (GPA 4.0/4.0)

Tsinghua University Beijing, China

B.S. in Math & Physics with High Honors