Software Engineer Platform and Inference

You will own the serving stack for frontier models, improving latency, throughput, GPU efficiency, batching, and autoscaling across a multi-cloud GPU fleet. You will build model pipelines, experiment and observability tooling, and operate 24/7 production systems.

Responsibilities

• Own the model-serving stack
• Improve latency, throughput, GPU efficiency, batching, and autoscaling
• Build product-ready model pipelines
• Build experiment and observability tooling
• Operate 24/7 systems with alerting and incident response
• Collaborate with researchers, product engineers, and the commercial team

Requirements

• 4+ years building production systems
• Depth in performance, distributed systems, or ML serving
• Experience optimizing model inference
• Experience with GPU utilization, batching, quantization, caching, or kernel-level work
• Experience owning 24/7 systems, including observability, alerting, and incident response
• Experience with 0-to-1 buildouts and 1-to-n scale-ups