Training Systems & AI Infrastructure Engineer

I work on large-scale training systems and AI infrastructure, focusing on distributed training, performance optimization, and reproducible workflows.

My interests span system-level tooling, backend integration, and benchmarking for large-scale model training.

• Large-scale model training systems
• Distributed & heterogeneous computing
• Performance benchmarking & diagnostics
• Training infrastructure tooling

• Languages: Python, C++, Bash
• Frameworks: PyTorch, Megatron-style training systems
• Systems: Distributed training, GPU performance, profiling
• Tooling: Benchmarking, preflight checks, reproducibility

• Unified training entry and tooling for multi-backend systems
• Training performance benchmarking and diagnostics
• Distributed training workflow design

📫 Contact

• Email: xmpeng.dev@gmail.com
• GitHub: https://github.com/xmpeng-dev