Xiaowei Jiang
Software Engineer at Anyscale
Palo Alto, California, United States
Summary
🤩
Rockstar🎓
Top SchoolXiaowei Jiang is a software engineer with four years of focused experience building scalable AI and distributed systems from Palo Alto. Currently at Anyscale, Xiaowei has contributed backend and MLOps improvements to Ray and added multimodal vision-language support to the high-throughput vLLM inference engine, highlighting a strength in productionizing ML infrastructure. Prior roles at Google and Uber reflect a solid track record in large-scale services and user-facing integrations, while an earlier hardware background at SanDisk and graduate work at UCLA bring cross-disciplinary systems and silicon-aware thinking. Known for tightening resource management, checkpointing, and CLI behavior in open-source projects, Xiaowei combines pragmatic engineering with an eye for reliability and performance. Fluent in both low-level hardware concerns and cloud-native ML tooling, they bridge model-serving features with rigorous distributed-systems practices.
4 years of coding experience
11 years of employment as a software developer
Bachelor's Degree, Physics, Bachelor's Degree, Physics at Peking UniversityUniversity of California, Los AngelesAdvanced Software System Graduate Certificate, Advanced Software System Graduate Certificate at Stanford University
English, Chinese
Github Skills (15)
llm10
computer-vision10
pytorch10
transformer10
machine-learning10
distributed-systems10
inference10
python10
testing10
api-design9
cicd9
mlops9
api9
apim9
tensorflow3
Programming languages (3)
Github contributions (5)
ray-project/ray
Jul 2021 - Jan 2023
Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
Role in this project:
Back-end Developer & MLOps Engineer
Back-end Developer & MLOps EngineerContributions:2 releases, 880 reviews, 178 commits in 1 year 6 months
Contributions summary:Xiaowei contributed to the Ray Tune library, implementing tests for dictionary flattening and fixing issues related to the CLI and how it interacts with the `--stop` flag. They also type-hinted the `TrialExecutor` and refactored and improved the handling of resources in a distributed environment, including updates on how the trial resources are handled during a resume operation. The user further introduced changes focused on improvements to checkpointing logic including file path sanitization, and also made improvements to the overall user experience by outputting informative messages related to insufficient resources.
pythonconsistsruntimetensorflowserving
vllm-project/vllm
Feb 2024 - Jul 2024
A high-throughput and memory-efficient inference and serving engine for LLMs
Role in this project:ML Engineer
ML EngineerContributions:63 reviews, 12 PRs, 89 comments in 5 months
Contributions summary:Xiaowei primarily contributed to the implementation and refinement of vision-language model (VLM) support within the vLLM project. This included adding features for multi-modal inputs, ensuring proper broadcasting of related data, and addressing associated bug fixes. The user's work focused on integrating VLM capabilities, particularly in the context of the OpenAI serving chat interface, and adjusting the model configurations to accommodate vision-related inputs.
amdcudagpthpuinference
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.