Sr. Simulation & Synthetic Data Engineer

Job description

About the Team Future Forward Research — Synthetic Data, builds the simulation and data infrastructure that powers Intuitive's autonomous surgical capabilities. The Role You'll build the virtual surgical worlds and the data pipelines that train our perception and policy models. Day to day, that means designing simulation environments, generating large volumes of labeled synthetic data, and working with ML engineers to close the sim-to-real gap for robotic surgery.

It's a hands-on engineering role at the intersection of 3D simulation, machine learning, and robotics.

What You'll Do

Design and build high-fidelity simulation environments for surgical tasks. Generate scalable synthetic datasets including photo realistic imagery, segmentation masks, depth, optical flow, and kinematic state — and own their quality, versioning, and delivery. Implement domain randomization and procedural scene generation to maximize sim-to-real transfer. Model deformable soft-tissue physics, tool-tissue contact, and instrument kinematics that match Intuitive's platforms.

Partner with ML engineers to define task curricula, reward functions, and evaluation benchmarks. Develop and refine sim-to-real transfer strategies, and validate simulated behaviors against benchtop phantoms and real systems. Build reusable tools, APIs, and documentation so the broader team can spin up new tasks without deep simulation expertise.

Qualifications

What You Bring We care more about real depth in a few of these than shallow coverage of all of them. Strong software engineering. You write maintainable, tested code in Python, C++ and/or C#, and you're comfortable in a Linux / Git / Docker / GPU workflow. Hands-on simulation or graphics experience. You've built things in a physics simulator, game engine, or rendering/VFX pipeline — robotics simulators, real-time engines, and offline graphics pipelines all count.

A feel for data and models. You've produced data that trained an ML model (or worked closely alongside that), and you understand how data quality and distribution show up in model behavior. Working knowledge of 3D and physics fundamentals — coordinate frames, rendering, rigid-body and contact dynamics — enough to reason about why a simulated scene does or doesn't look and behave correctly. You've likely spent several years building games, simulation, graphics, robotics, or data systems — but we're far more interested in what you've built than in a specific number on your résumé.

If you don't check every box, we still want to hear from you. Strong candidates often come from autonomous driving, humanoid robotics, gaming, or VFX rather than surgical robotics specifically.

Education

Bachelor's degree in Computer Science, Computer Graphics, Robotics, Electrical or Mechanical Engineering, Physics, or a related technical field — or equivalent practical experience. An advanced degree (MS or PhD) in a related area is a plus, not a requirement. We've hired strong engineers from all paths, including self-taught backgrounds and industry work in games, VFX, or robotics. Bonus Points None of these are required — they're signals, not gates.

Any one of them is a nice plus.

Deformable-object simulation

FEM, position-based dynamics, or differentiable physics NVIDIA Isaac Sim/Lab Surgical robotics simulation: ORBIT-Surgical, dVRK-based environments Sim-to-real techniques: domain randomization, system identification, privileged learning, residual policies Vision-language or vision-language-action models, and how simulation data supports them Procedural/generative asset creation: NeRFs, Gaussian Splatting, or diffusion models Distributed compute at scale: Ray, Kubernetes, Slurm, multi-GPU/multi-node Medical imaging or surgical video pipelines Publications or open-source work in simulation, graphics, robotics, or AI Tools You Might Use You won't touch all of these, and we don't expect you to walk in knowing them.

Category Examples Simulation engines Unity, NVIDIA Isaac Sim/Lab, MuJoCo Deformable physics PhysX FEM, SOFA, DiSECt, NVIDIA Warp/Newton, PBD Rendering & data USD, Omniverse Replicator, Blender Infrastructure Docker, Kubernetes, Ray, Airflow, CUDA What Success Looks Like in Your First Year Simulation environments for several core surgical tasks are operational and integrated with the ML training pipeline. Synthetic data delivers measurable model lift on targeted perception or policy metrics.

A quantitative sim-to-real correlation is established on benchtop phantoms. Parallelized infrastructure reliably generates thousands of training episodes per hour with automated dataset delivery. A reusable task-creation framework lets ML engineers define new surgical skills without deep simulation expertise. Shift Day Workplace Type Onsite - This job is fully onsite in the Sunnyvale campus.