Member of Technical Staff - ML Performance

📋 Description

• Build ML systems performant at scale
• Contribute to open-source projects and Modal's runtime
• Push language and diffusion models toward higher throughput
• Improve latency and efficiency of GPU inference
• Collaborate with research and engineering teams

🎯 Requirements

• 5+ years of high-quality, high-performance code
• Experience with torch, ML frameworks, and inference engines (vLLM or TensorRT)
• Familiarity with Nvidia GPU architecture and CUDA
• ML performance engineering: boost GPU perf, debug SM occupancy, reduce host overhead
• Nice-to-have: Linux kernel, file systems, containers