Senior Robotics Systems Engineer - Neural Reconstruction and Real2sim Applications

NVIDIA NVIDIA · Semiconductors · Santa Clara, CA

This role focuses on building end-to-end Real2Sim pipelines for robotics simulation, transforming sensor data into high-fidelity scene and object representations. It involves developing neural 3D reconstruction systems, integrating them with perception and navigation, and deploying these systems at scale for robotics applications.

What you'd actually do

  1. Build end‑to‑end Real2Sim pipelines that ingest multi‑modal sensor data (RGB/RGB‑D, LiDAR, IMU) and produce high‑fidelity scene and object representations for Isaac Sim.
  2. Develop neural 3D reconstruction systems and related 3D vision components to build digital twins of real‑world environments and objects.
  3. Integrate reconstruction outputs with mapping, localization, and multi‑sensor fusion systems for robust perception and navigation.
  4. Collaborate with ML researchers to integrate, deploy, and optimize workflows to enable training and validation in sim.
  5. Deploy and optimize reconstruction, data generation, and data augmentation pipelines at scale.

Skills

Required

  • BS, MS, or PhD in Computer Science, Electrical Engineering, Robotics, or related field (or equivalent experience).
  • 5+ years in robotics systems, 3D vision, simulation, or closely related software engineering roles.
  • Hands-on experience with neural 3D reconstruction, mesh/object reconstruction, or deep-learning–based 3D vision.
  • Exposure to reinforcement or imitation learning workflows for robotic perception and control.
  • Experience with ROS 2 and real‑time constraints, plus familiarity with Isaac Sim, Isaac Lab, or MuJoCo.
  • Proven ability to drive technical direction or architecture for complex robotics or 3D perception systems, with strong Python and C++ skills and experience shipping production‑quality systems.

Nice to have

  • Experience with neural scene representations (NeRF, 3D Gaussian Splatting, occupancy networks).
  • Experience with object reconstruction and sim‑ready asset generation pipelines (for example, mesh extraction, material and collision setup, USD export).
  • Familiarity with multi‑sensor fusion, sim‑to‑real transfer, and data augmentation workflows.
  • Experience building or integrating agentic frameworks, including LLM/VLM‑based task planners, tool‑use pipelines, or multi‑step reasoning systems.
  • Contributions to robotics or 3D vision open source, or publications at venues such as CVPR, ICCV, ICRA, RSS, or CoRL.

What the JD emphasized

  • shipping production-quality systems

Other signals

  • building dense 3D reconstruction systems
  • high-performance simulation workflows
  • robotics software that bridge research innovations with production-grade systems
  • transform sensor streams into structured assets and environment representations for downstream robotic applications