Ziniu Hu
Member Of Technical Staff at xAI
Palo Alto, California, United States
Summary
👤
Senior🎓
Top SchoolZiniu Hu is a research-driven machine learning engineer with eight years of experience bridging academic rigor and product-focused AI research, currently a Member of Technical Staff at xAI working on reinforcement learning and agentic coding. His PhD work in neural-symbolic methods and a string of internships and research roles at Google, DeepMind, Microsoft Research, and Caltech have produced papers and systems across LLM planning, retrieval-augmented multimodal pretraining, and billion-scale graph learning. He contributed to widely used open-source projects like pyHGT (Heterogeneous Graph Transformer) and has been recognized with multiple spot and peer bonuses for impactful collaborations. Comfortable moving models from theory to large-scale training and evaluation, he specializes in agent design and tool-augmented reasoning—skills now applied to reinforce Grok at xAI. An underappreciated thread through his work is his focus on practical robustness (e.g., hyperparameter and data-loading improvements) that consistently improves deployed model performance.
8 years of coding experience
1 year of employment as a software developer
Bachelor of Science - BS Computer Science, Bachelor of Science - BS Computer Science at Peking UniversityUniversity of California, Los Angeles
Github Skills (6)
transformer10
pytorch10
machine-learning10
graph-neural-network10
python10
nlp7
Programming languages (2)
Github contributions (5)
acbull/pyHGT
Feb 2020 - Jun 2022
Code for "Heterogeneous Graph Transformer" (WWW'20), which is based on pytorch_geometric
Role in this project:
Data Scientist
Data ScientistContributions:173 commits, 11 PRs, 149 pushes in 2 years 4 months
Contributions summary:Ziniu primarily worked on training and evaluating models for the author disambiguation and paper classification tasks. Their commits include modifications to training scripts, such as updates to hyperparameters like dropout and the addition of gradient clipping. Furthermore, the user made changes to the model and data loading procedures, improving the overall performance of the machine learning models.
pytorchheterogeneousgeometric-deep-learninggnnnetwork-embedding
acbull/acbull.github.io
Sep 2017 - Dec 2022
Contributions:183 commits, 414 pushes, 15 branches in 5 years 4 months
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.