Software Engineer, Inference (AI Data Engineering)

About this role

SpaceX seeks a Software Engineer to design and optimize high-performance AI inference platforms that serve mission-critical models across the organization. You'll own distributed infrastructure, low-level GPU optimizations, and end-to-end serving systems supporting SpaceX's most ambitious engineering goals.

What you'll do

• Design and build highly reliable, high-throughput inference systems serving AI models across SpaceX
• Architect scalable distributed infrastructure including load balancing, auto-scaling, batch scheduling, and KV cache management
• Optimize inference latency and throughput via GPU kernels, quantization, speculative decoding, and other acceleration techniques
• Develop mission-critical serving systems with 100% uptime, low tail latency, and strong observability
• Benchmark and accelerate inference engines like SGLang, vLLM, and TensorRT-LLM for production workloads
• Own end-to-end components including request routing, SDK development, rate limiting, and CI/CD infrastructure

What they're looking for

• Distributed systems design and implementation
• Rust or C++ for systems programming
• GPU optimization and kernel programming
• LLM inference engines (vLLM, SGLang, TensorRT-LLM, Triton)
• High-concurrency production serving systems
• Backend development and full-stack engineering
• Service observability and reliability practices
• Quantization, batching, caching, and parallelism