Jobs at UMA

This open candidate position invites individuals passionate about Robotics and AI to propose innovative ideas and solutions in a fast-paced startup environment.

Own the whole-body controller and state estimation for a humanoid robot on real hardware. Responsibilities include designing locomotion and loco-manipulation controllers, floating-base estimation, contact detection, sensor fusion, real-time optimal control, actuator characterization, simulation-to-hardware deployment, model-based and learned policy integration, and diagnostic tooling.

Develop and implement learning-based control systems for humanoid robots, using reinforcement learning and other techniques, while optimizing performance and conducting experiments.

Own and scale UMA’s compute infrastructure for large-scale AI training, evaluation, and data-processing workloads. Responsibilities include multi-provider GPU capacity, cloud scheduling, distributed training frameworks, checkpointing, preemption, virtualized development environments, observability, reliability, cost optimization, and provider relationships. The role requires hands-on operation of production-grade GPU clusters and may grow into leadership of the compute infrastructure team.

Implement and reproduce state-of-the-art world-model baselines, build training/data/evaluation pipelines, run latent-space planning and MPC experiments in simulation and on humanoid robots, and scale and debug distributed GPU training to keep experiments reproducible and logged.

Owns finance, legal, and administrative operations for a deep-tech startup. Responsibilities include fundraising support, investor reporting, cash forecasting, budgeting, grant and tax incentive applications, cap table and contract administration, supplier payments, banking, quarterly close, and finance automation. The role partners closely with the CFO, founders, external accountants, counsel, and DevOps while building scalable processes in a fast-growing environment.

Lead research on world models, planning, and reasoning for humanoid robots. Design and run experiments in simulation and on hardware, own data processing, model training, evaluation, and integration with control stacks, and document and mentor engineers to ship real-world capabilities.