Build AI
ML Engineer, Inference Optimization
$200K–$320K
San Francisco · On-site · Full-time · Posted Aug 30, 2026
- Experience
- Real inference optimization experience required; no years specified.
About the job
Optimizes model inference to reduce latency and compute costs while maintaining sufficient quality to scale Build AI’s data engine and product.
What you'll do
- Own inference latency, throughput, and cost-per-work performance; identify bottlenecks through profiling and ship fixes.
- Optimize kernels, batching, quantization, compilation, serving, and hardware utilization.
- Build serving and evaluation workflows that make inference costs visible, and collaborate with research and product on affordable model deployment.
What they ask for
- Strong ML or systems engineering skills with real inference optimization experience.
- Proficiency in Python and in C++ or Rust for performance-critical paths.
- Familiarity with PyTorch or JAX and profiling tools.
- Comfort working in a small research team under cost pressure.
Nice to have
- Production experience with CUDA, kernels, compilers (TVM, MLIR, TensorRT), or quantization.
- Experience owning GPU or accelerator costs, serving video or large models, or working with memory hierarchy, data movement, and low-precision compute.
Pay
$200K – $320K (base salary)
Tools
Python, C++, Rust, PyTorch, JAX, CUDA, TVM, MLIR
AI focus
Optimizes inference for AI models, including serving and evaluation, with emphasis on latency, throughput, and cost at scale.
Model-written summary · the original posting is authoritative