Design, implement, and operate distributed training, mixed-precision, gradient-checkpointing, and GPU-kernel optimizations for VLM, VLA, and foundation-model training.
Develop and optimize reinforcement-learning frameworks and performance for driving foundation models.
Profile and debug training-throughput bottlenecks across compute, compiler, framework, and cluster-networking layers.
Collaborate with ML researchers, data engineers, and hardware engineers on requirements and architecture for large-scale training infrastructure.
Implement custom kernels and build benchmarks to validate training-performance optimizations.
Requirements
Require 4+ years of total experience, including 2+ years working on large-scale model training platforms.
Experience with Megatron-LM, TorchTitan, and large-scale GPU or accelerator clusters.
Experience with RL frameworks such as Verl or MILES and RL algorithms including GRPO and PPO.
Experience with NCCL and optimizing communication/compute overlap in multi-node training.
Experience using NVIDIA Nsight, PyTorch Profiler, or XLA/JAX profiling tools to identify and resolve training bottlenecks.
Proficiency in Python or C++.
Experience with ML compilers, performance tooling, or hardware/software co-design, including GPU, TPU, or accelerator architectures.
Salary: $177k - $270k/yr
Zoox
Zoox is transforming mobility-as-a-service by developing a fully autonomous, purpose-built fleet designed for AI to drive and humans to enjoy.