Staff Simulation Engineer - Core Simulation Platform

Apptronik Apptronik · Robotics · Mountain View, CA · Software Engineering

Staff Simulation Engineer for Core Simulation Platform at a human-centered robotics company. The role focuses on building and scaling a high-performance simulation platform for RL training, controls validation, and CI/CD integration testing. Responsibilities include contributing to platform architecture, developing cloud-native pipelines for millions of experience hours per day, improving sim-to-real transfer, developing sensor-realistic environments, and shipping platform improvements to reduce iteration time for RL, controls, and perception teams.

What you'd actually do

  1. Contribute to the architecture and development of the core simulation platform — high-performance digital twins that serve as the foundation for policy training, controls validation, and CI/CD integration testing.
  2. Help build and scale cloud-native simulation pipelines capable of generating millions of experience hours per day, parallelizing physics and rendering for rapid policy iteration.
  3. Improve sim-to-real transfer through advanced contact models and actuator dynamics that help policies trained in simulation transfer to physical humanoid hardware.
  4. Develop sensor-realistic environments (camera, LiDAR, depth) that challenge the perception stack with dynamic and diverse worlds.
  5. Ship platform improvements that visibly reduce iteration time for RL, controls, and perception teams.

Skills

Required

  • Excellent C++ and/or Python programming skills
  • Familiar with software development best practices: CI/CD, automated testing, and code quality standards
  • Understanding of robotics concepts: kinematics, dynamics, controls, system identification
  • Experience with modern robotic simulators (Isaac Lab, MuJoCo, or equivalent)
  • Understanding of Reinforcement Learning with research or production experience
  • Experience with large-scale training workloads — deploying parallel simulations on cloud platforms (AWS, GCP, Azure) with distributed computing frameworks (e.g., Ray, Kubernetes)

What the JD emphasized

  • fast enough for RL training
  • accurate enough for controls validation
  • stable enough to be the team’s daily driver
  • high-performance digital twins
  • cloud-native simulation pipelines
  • millions of experience hours per day
  • rapid policy iteration
  • sim-to-real transfer
  • physical humanoid hardware
  • large-scale training workloads
  • parallel simulations on cloud platforms
  • distributed computing frameworks

Other signals

  • simulation platform for RL training
  • large-scale training workloads
  • sim-to-real transfer
  • cloud-native simulation pipelines