Staff ML Engineer, VLA Training at Nidus

As a Staff ML Engineer at Nidus, you will lead the development of vision-language-action (VLA) models that power intelligent robots. You’ll own the entire lifecycle—from dataset construction and distributed training on multi-node GPU clusters to on-robot validation. Your work directly enables robots to perceive, understand, and act in dynamic manufacturing environments.

• Train and optimize large-scale vision-language-action models using PyTorch or JAX, implementing architectures like imitation learning, behavior cloning, and diffusion policies.
• Architect and operate distributed training pipelines across multi-node GPU clusters, focusing on improving throughput, stability, and utilization for heterogeneous robot datasets.
• Partner with robotics and hardware teams to run structured experiments on physical robots, translating training results into reliable autonomous capabilities and task success.

• Five or more years of professional experience in machine learning or robotics, with a proven track record of training modern deep learning models at scale.
• Expert-level proficiency in building distributed training pipelines and debugging complex systems across input data, training dynamics, and real-world robot behavior.
• U.S. citizenship is required for this role due to the nature of our work and customers; must be willing to work in-person in New York City.

• High-impact ownership as a Staff IC, leading the modeling and infrastructure strategy for the core intelligence layer of a seed-stage robotics company.
• Exceptional traction with 7-figure revenue already closed from tier 1 prime customers since inception earlier this year, signaling strong market fit.
• Opportunity to work in-person in NYC with an elite team from defense and AI backgrounds, directly iterating on physical hardware and cutting-edge VLA models.