Principal Engineer, Model Performance

Highlights

• Own how fast and cheaply every model runs across the whole fleet.
• Serve new open models at the Pareto frontier within days of release.
• Power inference for 20B+ requests across 400,000+ models worldwide.

What they offer

• Meaningful stock options
• Generous paid time off including vacation, sick days and public holidays
• Paid maternity, paternity and caregiver leave
• Twice-yearly company retreats in inspiring locations
• Remote-first with core collaboration hours and flexible schedule

Ideal candidate

A hands-on performance leader who still writes kernels and has driven frontier inference gains into production. You obsess over latency and cost, set bold technical roadmaps, and mentor small elite teams to deliver without you.

Requirements

• Led inference performance work at the frontier
• Hands-on CUDA kernel development and tuning
• Experience with quantization, parallelism and batching for production inference
• Experience owning benchmarking and performance regression systems
• Experience hiring, mentoring and leading senior engineers