Xiaohan Zhang
Research Engineer at Databricks Mosaic Research
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolXiaohan Zhang is a research engineer blending 8+ years of experience in ML/AI productization with a deep background in computational mechanics and HPC from a PhD at Carnegie Mellon. He has built and deployed production-scale models for ranking, personalization, natural language search, forecasting and content moderation at Salesforce and now accelerates generative AI training and streaming data infrastructure at Databricks Mosaic Research. His work spans from GPU- and MPI-accelerated C++ simulators for materials science to end-to-end ML systems and optimized data pipelines (notably contributing GCS support to MosaicML's Composer and Delta-to-JSON tooling for LLM training). Equally comfortable in research and engineering contexts, he leverages physics-informed modeling and large-scale systems thinking to solve practical ML problems for enterprise customers. Based in Mountain View, he pairs academic rigor with a track record of shipping reliable, high-performance software for both science and industry.
8 years of coding experience
8 years of employment as a software developer
BS Hydraulic engineering, BS Hydraulic engineering at China Agricultural UniversityMaster of Engineering (MEng) Geotechnical Engineering mechanics, Master of Engineering (MEng) Geotechnical Engineering mechanics at Tsinghua UniversityCarnegie Mellon University
Github Skills (20)
json10
pytorch10
delta-lake10
data-pipelines10
python10
machine-learning10
pyspark10
google-cloud-storage10
deep-learning10
neural-network10
data-pipeline10
pyarrow9
dockers9
llm9
docker9
Programming languages (1)
Github contributions (5)
mosaicml/llm-foundry
Nov 2023 - Nov 2024
LLM training code for Databricks foundation models
Role in this project:
Back-end Developer & Data Engineer
Back-end Developer & Data EngineerContributions:71 reviews, 62 PRs, 78 pushes in 1 year
Contributions summary:Xiaohan primarily worked on integrating and optimizing data pipelines for a Delta table within a Databricks environment. They focused on reading data from a Delta table using `databricks-sql` and converting it to JSON format. Their contributions involved optimizing the data conversion process, adding features like compression and arrow support, and integrating the workspace client. They also implemented testing and addressed linting issues.
deep-learningllmneural-networksnlppytorch
mosaicml/composer
Jun 2023 - Nov 2024
Supercharge Your Model Training
Role in this project:ML Engineer
ML EngineerContributions:1 release, 13 reviews, 14 PRs in 1 year 4 months
Contributions summary:Xiaohan primarily contributed to integrating and updating the Google Cloud Storage (GCS) backend within the composer repository, replacing the existing libcloud implementation. They implemented GCS support, wrote tests, and refactored code. Additionally, the user addressed versioning issues by pinning and restoring specific package versions to resolve code quality failures. The user also fixed a typo in the trainer documentation.
pytorchml-systemsdeep-learningneural-networksmachine-learning
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.