Machine Learning: Whole-Body Control

The Bot Company; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

The Bot Company is building high-performance whole-body controllers that enable robust, agile motion and manipulation in the real world.

You will train low-level control policies in simulation and own the stack from environment design to large-scale training and sim-to-real deployment.

You’ll join a small elite team where speed, deep engineering instinct, and systems thinking drive rapid, measurable results.

Qualifications

• Very strong programming in Python, C++, or Rust; CUDA is a plus.
• Strong foundations in modern reinforcement learning.
• Experience training embodied control policies.
• Hands-on robotics simulation experience.
• Solid intuition for dynamics, contact, and actuation.
• Comfortable running large-scale GPU experiments.

Responsibilities

• Train whole-body policies for locomotion, manipulation, and coordinated motion.
• Build scalable simulation environments in IsaacLab, MuJoCo, or similar with parallel rollouts.
• Design rewards and curricula for stable long-horizon learning.
• Own sim-to-real transfer using domain randomization and structured evaluation.
• Run and debug large-scale GPU training experiments.

Skills

Python

C++

Rust

CUDA

Reinforcement learning

Robotics simulation

Job description

What we look for in all candidates

• Exceptional mental acuity: you think quickly, learn instantly, and reason across unfamiliar domains.
• Engineering curiosity: you naturally dig into how systems work, even outside your specialty.
• High performance mindset: you move fast, handle ambiguity, and excel when the environment is demanding.

Machine Learning: Whole-Body Control

We are building high-performance whole-body controllers that produce robust, agile motion and manipulation in the real world.

You will train low-level control policies in simulation and own the stack from environment design to large-scale training and sim-to-real deployment.

What You'll Do

• Train whole-body policies for locomotion, manipulation, and coordinated motion.
• Build scalable simulation environments in IsaacLab, MuJoCo, or similar with parallel rollouts.
• Design rewards and curricula that enable stable long-horizon learning.
• Own sim-to-real transfer using domain randomization and structured evaluation.
• Run and debug large-scale GPU training experiments.

Requirements

• Very strong coding skills in Python, C++, or Rust. CUDA is a plus.
• Strong foundations in modern reinforcement learning.
• Experience training embodied control policies.
• Hands‑on robotics simulation experience.
• Solid intuition for dynamics, contact, and actuation.
• Comfortable running large-scale GPU experiments.

Why Join

You’ll work with a small, elite team on challenges that require speed, intelligence, and deep engineering instinct. If you enjoy understanding systems at all levels, move fast, and think even faster, you’ll thrive here.