SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 175,000 - 400,000 / annual
MatX is building custom chips optimized for training and serving large language models, spanning silicon design through compilers and kernels. The company is hiring compiler engineers to design and implement toolchains that compile Python eDSL into executables for proprietary hardware, primarily using Rust.
In this role, you will focus on mid-end and backend compiler work, including program analysis, optimization, instruction selection, scheduling, and register allocation. You'll collaborate closely with AI researchers and hardware architects to achieve optimal performance on MatX's proprietary hardware, translating high-level ML models into efficient low-level code.
Key responsibilities include designing compiler passes for data flow analysis, register allocation, instruction scheduling, and loop optimizations. You'll work on the full compilation pipeline, from Python-level abstractions down to hardware-specific optimizations, ensuring best-in-class latency and throughput per dollar and per watt.
You should have strong software engineering fundamentals with deep expertise in compiler algorithms and low-level optimization. Experience with LLVM, MLIR, SIMD, NVPTX, or XLA is highly valued. ML experience—such as familiarity with Transformers, GPU kernels, or ML optimization—is a plus. You'll thrive in a dynamic environment where upstream and downstream changes require flexibility and proactive problem-solving.
This is a hybrid role requiring 3 days per week in the Mountain View office (Tuesday–Thursday), with flexibility to work remotely on Mondays and Fridays.
