Luminal
Making AI run fast on any hardware.
ML Framework Frontend Engineer
$125K - $250K•0.25% - 1.00%•San Francisco, CA, US
Job type
Full-timeRole
Engineering, Machine learningExperience
Any (new grads ok)Visa
Will sponsorSkills
Torch/PyTorchConnect directly with founders of the best YC-funded startups.
Apply to role ›
Jake Stevens
Founder
Jake Stevens
Founder
About the role
What you’ll do
- Build a custom
torch.compilebackend that lowers PyTorch graphs into Luminal’s IR - Own framework interop edge cases: graph breaks, dynamic shapes, weird ops, numerics
- Build a correctness + testing suite for CUDA and the other ML stacks we touch
- Ship benchmarking + visualization tooling so users can trust the speedups
- Wrap kernel libs like CUTLASS so they slot cleanly into our IR
- Talk to the Luminal community (Slack/GitHub/calls), figure out what’s painful, and ship fixes fast
You should probably have
- Strong Python + deep PyTorch (torch.compile / Dynamo / FX-ish)
- Comfortable with CUDA/profiling/perf debugging
- Good taste for APIs + devtools
- Bonus: Rust, CUTLASS, compiler/IR experience
- Huge bonus: you’ve contributed to an open source ML framework before
About Luminal
Luminal (YC S25) is focused on optimizing AI models to accelerate and simplify model deployment. We are building an AI compiler that enhances model speeds by 10x and streamlines deployment to production with just a single line of code. Our mission is to turn AI research code into production code, seamlessly.
Role Description
This is a full-time on-site role for a Founding Engineer located in downtown San Francisco. The Founding Engineer will be responsible for assisting the design of the core compiler. Day-to-day tasks will include writing CUDA kernels, conducting model performance reviews, and shitposting on social media.
Founders
