As a member of the Dojo group, you may be chargeable for permitting Teslas neural networks to educate effectively on our upcoming in-residence custom-silicon supercomputer systems. Join a small group of skilled builders in automating the compilation of PyTorch-derived neural community graphs into applications that run on Teslas custom vastly parallel Dojo accelerators. The perfect candidate is an initiative-taking, proactive engineer to make certain the Dojo software program excellent is excessive and the personal experience is seamless.
What Youll Do - Developing Python libraries to run, monitor, measure, and troubleshoot deep mastering workflows strolling on Dojo.
- The position may be programming about 50% of the time in Python. The last time may be spent debugging, experimenting, and investigating.
- Writing checks to assure correctness at each stage of the Dojo stack, from excessive-stage pytorch integration, thru each degree of the ML compiler stack, right all the way down to our custom hardware.
- Ensuring that neural networks of hobby to our customers characteristic efficaciously on Dojo on the anticipated excessive overall performance.
- Writing gear to run, monitor, measure, and troubleshoot deep mastering workflows strolling on Dojo.
- Writing bench-marking and reporting gear.
- Supporting in-residence customers; triaging mistakes and overall performance bottlenecks, using problems to root cause, and supplying workarounds and fixes.
- Anticipating probable use cases; fixing issues earlier than our customers hit them.
What Youll Bring - BS in Computer Science, Physics, Computer Engineering, or Electrical Engineering, evidence of exquisite competencies in associated fields with sensible software program engineering experience, or equivalent
- Strong proficiency with Python and snug with C++
- Highly acquainted with Linux management and internals
Nice to Have: - Deep-mastering frameworks (PyTorch)
Apply!