Ray Summit 2025, November 2025
Boosting vLLM Inference on Huawei NPU with Ray Compiled Graphs
A Ray Compiled Graphs extension with an NPU backend for faster vLLM inference on Ascend NPUs.
Talks
Talks, tutorials and recorded presentations.
Featured talks
Ray Summit 2024, October 2024
Scaling Ray to 10K NPUs: Huawei's Hyperscale Journey
How our team runs Ray on clusters of 10,000 Huawei Ascend NPUs.
All talks 14
Software Engineering for Foundation Models (SE4FM)
ICSE 2026. Co-presenter
Turbocharging AIware: Performance Engineering for Production-Ready FMware
CASCON 2025. Co-author
Boosting vLLM Inference on Huawei NPU with Ray Compiled Graphs
Ray Summit 2025, San Francisco, CA. Speaker, with Zhilong Chen and FengChun Hua
Performance Engineering for AIware
ICSE 2025, AIware Bootcamp Mini, Ottawa, Canada. Co-presenter, with Haoxiang Zhang
Software Performance Engineering for FMware
AIware Leadership Bootcamp 2024, Toronto, Canada. Speaker, with Haoxiang Zhang
Scaling Ray to 10K NPUs: Huawei's Hyperscale Journey
Ray Summit 2024, San Francisco, CA. Speaker, with Chong Yin Tan and Xiaoshuang Liu
Towards Build Verifiability for Java-Based Systems
ICSE-SEIP 2022, Online. Presenter
Towards Training Reproducible Deep Learning Models
ICSE 2022, Online. Speaker
Studying the Use of Java Logging Utilities in the Wild
ICSE 2020, Online. Speaker
An Industrial Experience Report on Performance-Aware Refactoring on a Database-Centric Web Application
ASE 2019, San Diego, CA. Speaker
Extracting and Studying the Logging-Code-Issue-Introducing Changes in Java-Based Large-Scale Open Source Software Systems
ASE 2019, San Diego, CA. Speaker
Improving the Software Logging Practices in DevOps
ICSE 2019 Doctoral Symposium, Montréal, Canada. Speaker
An Automated Approach to Estimating Code Coverage Measures via Execution Logs
ASE 2018, Montpellier, France. Speaker
Characterizing and Detecting Anti-Patterns in the Logging Code
ICSE 2017, Buenos Aires, Argentina. Speaker