Xuechen Li
Member Of Technical Staff at xAI
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolXuechen Li is a Member of Technical Staff at xAI and a Stanford PhD candidate in Computer Science with a decade of experience building and evaluating machine learning systems. His background spans Google AI Residency and research roles, a Microsoft internship on differentially private ML, and early work in approximate Bayesian inference, giving him strong foundations in both theory and applied systems. He contributes to open-source projects such as google-research/torchsde (GPU-enabled differentiable SDE solvers), stanford_alpaca (training and weight-diff tooling), and HELM (copyright and data-extraction metrics), reflecting fluency from low-level library fixes to model-weight surgery and evaluation metrics. Xuechen blends product-driven post-training and large-scale RL research with hands-on engineering—comfortable patching core dependencies and Brownian process handling while driving reproducible model training and privacy-aware evaluation.
10 years of coding experience
2 years of employment as a software developer
Beijing No.4 High SchoolBachelor of Science - BS (Hons), Bachelor of Science - BS (Hons) at University of TorontoDoctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Stanford University
Chinese, English, French
Github Skills (20)
transformers10
pytorch10
python10
nltk10
machine-learning10
differential-equations10
deep-learning10
trainings10
modeling10
data-science9
language-models9
language-model9
deep-neural-networks9
neural-networks9
neural-network9
Programming languages (8)
Github contributions (5)
google-research/torchsde
Jul 2020 - May 2021
Differentiable SDE solvers with GPU support and efficient sensitivity analysis.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:4 releases, 93 reviews, 101 commits in 10 months
Contributions summary:Xuechen replaced placeholder packages and fixed import issues, indicating involvement in project setup and dependency management. They also addressed brownian import issues by changing relevant code, indicating a focus on core libraries. Further, they added functionality to handle "alternative color" and "to device" calls for a brownian tree, showing engagement with internal structures and potential refactoring efforts.
cudastochastic-differential-equationsdifferentiablesensitivity-analysisgpu
tatsu-lab/stanford_alpaca
Mar 2023 - Jun 2023
Code and documentation to train Stanford's Alpaca models, and generate the data.
Role in this project:ML Engineer
ML EngineerContributions:13 PRs, 10 pushes, 10 branches in 3 months
Contributions summary:Xuechen primarily contributed to the training and weight management aspects of the Alpaca model. Their commits involved modifying the `train.py` script, indicating work on training code and related configurations. Furthermore, the user introduced and refined the `weight_diff.py` script, demonstrating an understanding of weight difference calculations and model recovery. This suggests the user focused on model training, optimization, and the handling of model weights.
deep-learninginstruction-followinglanguage-model
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.