Dacheng Li
Doctoral Student at Sailing Lab
Berkeley, California, United States
Summary
🤩
Rockstar🎓
Top SchoolDacheng Li is a PhD student at UC Berkeley specializing in machine learning and distributed systems with eight years of research and engineering experience across academia and startups. His work—mentored by Ion Stoica and Joseph Gonzalez—focuses on scaling long-context LLM training and human-evaluated LLM benchmarks, with publications including DISTFLASHATTN (COLM 2024), Chatbot Arena (ICML 2024), and NeurIPS 2023 analyses. Previously he contributed to MPC-friendly transformer inference and automated model-parallel strategy search (ICLR/NIPS), and co-founded an AI-driven housing recommender, blending systems thinking with product impact. Trained at Carnegie Mellon (MS, Machine Learning) and UCSD (BS in CS & Math), he combines deep theoretical understanding with practical system-building for large-scale ML workloads. An interesting through-line: he repeatedly tackles privacy, efficiency, and real-world evaluation together—bringing human preference signals into systems designed for training and inference at scale.
8 years of coding experience
3 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of California, BerkeleyBachelor's degree, Computer Science, Mathematics, Bachelor's degree, Computer Science, Mathematics at 美国加州大学圣迭戈分校Master's degree, Machine Learning, Master's degree, Machine Learning at 美国卡内基梅隆大学Master of Science - MS, Machine Learning, Master of Science - MS, Machine Learning at Carnegie Mellon University
Github Skills (83)
parallel10
python10
hyperparameter-optimization10
deep-learning10
gpu10
lora10
ray10
gpu-acceleration10
optimization10
numpy10
llm-inference10
ai10
neural-network10
tensor10
chatbot10
Programming languages (6)
Github contributions (5)
DachengLi1/AMP
Oct 2022 - Nov 2022
(NeurIPS 2022) Automatically finding good model-parallel strategies, especially for complex models and clusters.
Contributions:35 commits, 28 pushes, 3 branches in 1 month
DachengLi1/MPCFormer
Aug 2022 - Jan 2023
(ICLR 2023 Spotlight) MPCFormer: fast, performant, and private transformer inference with MPC
Contributions:33 commits, 7 pushes, 13 comments in 4 months
inference
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.