Projects
Leading projects in AI4Science and ML infrastructure
MindScience Platform
Lead maintainer of Huawei's flagship AI4Science platform, enabling breakthrough scientific computing across Earth sciences, fluid dynamics, electromagnetics, chemistry, and biology. Features 6 specialized suites (MindEarth, MindFlow, MindElec, MindChemistry, MindSPONGE, SciAI) with 60+ AI models, achieving 1000x performance improvements over traditional methods in weather forecasting and CFD simulations.
Key Features:
- 6 specialized scientific computing suites
- 60+ AI models for various domains
- 1000x performance improvements
Specialized Suites:
MindEarth - Weather Forecasting
Earth sciences suite supporting multi-temporal-spatial-scale meteorological forecasting with 1000x performance improvement over traditional methods.
- Medium-range weather forecasting
- Precipitation nowcasting models
MindFlow - Computational Fluid Dynamics
CFD solver suite supporting AI fluid simulation driven by physics, data, and data-mechanism fusion with built-in differentiable CFD solver.
- Physics-informed fluid simulation
- PHengLei CFD solver integration
MindSPONGE - Computational Biology
Computational biology toolkit supporting high-performance, modular, end-to-end differentiable, AI-native architecture for molecular simulation and protein folding.
- Molecular simulation software
- MEGA-Protein structure prediction
ML Infrastructure
LLM Serving Research
Specialized research in large language model serving optimization with expertise in vLLM and vLLM-Ascend systems. Focus on performance optimization, distributed inference, and ML infrastructure for production-scale deployments.
Key Features:
- vLLM optimization and contributions
- vLLM-Ascend integration and performance
- Distributed inference systems