SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Zoox is building autonomous robotaxis from the ground up and is seeking a Software Engineer to optimize ML training performance. The ML Platform team enables large-scale foundation models, vision language models (VLMs), and vision language action models (VLAs) to power autonomous driving systems.
In this role, you will design, implement, and operate cutting-edge ML training performance optimization techniques including distributed training (data/tensor/pipeline parallelism), mixed-precision training, gradient checkpointing, and custom GPU kernels to scale VLM, VLA, and foundational model training efficiency. You will work on reinforcement learning frameworks and RL performance optimization for driving foundation models, profile and debug training throughput bottlenecks across compute, compiler, framework, and cluster-networking stacks, and collaborate closely with ML researchers, data engineers, and hardware engineers to define requirements and align on architectural decisions for large-scale training infrastructure. You will implement custom kernels and build benchmarks to track and validate training-performance optimizations across the full stack.
This is an opportunity to work with state-of-the-art accelerators and cutting-edge techniques, directly supporting Zoox's Perception, Prediction, Planner, Simulation, and Collision Avoidance teams. You will act as a force multiplier for internal customers and have significant growth opportunities as Zoox expands robotaxi deployments and ventures into new ML domains.
REQUIREMENTS:
- 4+ years of total experience, including 2+ years working on large-scale model training platforms
- Experience with distributed training frameworks (Megatron-LM, TorchTitan) and efficiently utilizing GPU/accelerator clusters at scale
- Experience with RL frameworks such as Verl/MILES and RL algorithms (GRPO, PPO)
- Experience with collective communication libraries (e.g., NCCL) and optimizing communication/compute overlap in multi-node training
- Experience using profiling tools (NVIDIA Nsight, PyTorch Profiler, or XLA/JAX profiling) to identify and resolve training bottlenecks
- Proficiency in Python or C++
- Experience with ML compilers, performance tooling, or hardware/software co-design (e.g., GPU/TPU/accelerator architectures)
Note: You do not need to meet all requirements to be considered.
