Software Engineer, Performance, AI Infrastructure

Job Description

As a Software Engineer within the AI group, you'll optimize and scale neural network training and auto-labeling infrastructure for Autopilot and the Humanoid robot. Your role will focus on reducing training job times, integrating low-level code with high-level frameworks, profiling workloads, and optimizing for hardware utilization, including CPU and GPU resources.

Requirements

• Extensive experience in CUDA kernel programming
• Experience in Python
• Experience with deep learning frameworks (ideally PyTorch)
• Experience in profiling CPU/GPU code
• Proficient in hardware-software interactions and resource utilization
• Good knowledge of CUDA kernels for deep learning
• Experience with high-performance networking (e.g., Infiniband, RDMA, NCCL)
• Experience with Triton preferred.

Responsibilities

• Reduce wall clock time for training jobs
• Integrate efficient low-level code
• Profile workloads to increase efficiency
• Optimize for hardware utilization.

Benefits

• Benefits include Aetna PPO and HSA plans with $0 payroll deduction, family-building benefits, dental and vision plans with $0 contribution, flexible spending accounts, 401(k) with employer match, company-paid life insurance, sick and vacation time, and employee discounts.