ML Engineer, Inference Optimization

Job Summary

Inference is about 90% of compute spend. Economics are heavily driven by inference optimization. We’re hiring someone to make inference cheaper, faster, and good enough that we can scale the data engine and the product without the GPU bill eating the company.

Key Responsibilities

• Own inference performance: latency, throughput, and cost per unit of work (tokens, frames, or jobs)
• Cut the 90% compute line: kernels, batching, quantization, compilation, serving, and hardware utilization
• Profile pipelines (Nsight, PyTorch Profiler, or equivalent), find the real bottleneck, and ship the fix
• Work with research and product so models that are accurate are also affordable to run at scale
• Build the serving and eval path so experiments don’t hide the inference bill
• Measure cost as a first-class metric, not an afterthought once quality is “done”

You may be a good fit if you have (Must-have qualifications)

• Strong ML / systems engineer with real inference optimization experience (serving, compilers, CUDA/kernels, quantization, or similar)
• Comfortable in Python and in C++ or Rust for performance-critical paths
• You think in dollars and tokens/frames per second, not only in accuracy tables
• Familiarity with PyTorch (or JAX) and with profiling tools
• Comfortable in a small research team shipping under cost pressure

Strong candidates may also have experience with (Nice-to-have qualifications)

• CUDA, kernels, compilers (TVM, MLIR, TensorRT), or quantization in production
• You have owned GPU/accelerator cost as a first-class metric
• Serving stacks for video or large models
• Understanding of memory hierarchy, data movement, and low-precision compute

Benefits

• Competitive pay
• Medical, dental, and vision packages with generous premium coverage
• $500 per month credit for waiving medical benefits
• Housing subsidy of $2k per month for those living within walking distance of the office
• Relocation support for those moving to San Francisco (Financial District) or Shenzhen (Nanshan)
• Various wellness benefits covering fitness, mental health, and more
• Daily lunch and dinner in our office
• Unlimited compute budget subject to ROI justification
• Unlimited Codex and Claude credits
• Travel

How we're different

We are a fully in-person team in San Francisco (Financial District) and Shenzhen (Nanshan), and greatly value engineering skills. We do not have boundaries between engineering and research, and we expect all of our technical staff to contribute to both and work across disciplines as needed.