Robotics Simulation Expert

Dreamvu; Rangareddy; On-site; INR 1,800,000 - 2,600,000; Full time

Job summary

Dreamvu seeks a Robotics Simulation Expert to own and evolve the simulation pipeline for robot learning and testing, intersecting physics simulation, 3D content pipelines, and ML to build high-fidelity digital twins for training and sim-to-real transfer.

You’ll work with NVIDIA Isaac Sim, Newton physics engine, USD, Blender, and Gaussian Splatting to create environments and assets enabling robust learning and validation of robotic systems.

Qualifications

• Master’s degree in Machine Learning, Robotics, CS, or related field.
• Hands-on experience with NVIDIA Isaac Sim for robotics simulation.
• Strong knowledge of USD internals and 3D asset pipelines; Blender experience.
• Experience with Gaussian Splatting for 3D reconstruction is desired.

Responsibilities

• Build and maintain robot simulation environments in Isaac Sim.
• Tune Newton physics engine for realistic dynamics.
• Develop pipelines converting Gaussian Splat reconstructions into assets.
• Own end-to-end asset pipeline from capture to import into Isaac Sim.
• Work with USD stages, layers, and references to optimize scenes.
• Collaborate with ML/perception teams to validate simulation accuracy.
• Document pipelines and best practices.

Skills

NVIDIA Isaac Sim

Gaussian Splatting

Blender

USD internals

Python/C++

ROS/ROS2 bridges

Machine Learning in simulation

Gazebo knowledge

Education

Master’s degree in Machine Learning, Robotics, Computer Science

Tools

Newton physics engine

USD

Gazebo

ROS/ROS2

Job description

Robotics Simulation Expert

Location: [Hyderabad / work from office] Department: Robotics / Simulation & ML Reports to: [Lead, Robotics Simulation / Engineering Manager] Employment Type: Full-time

About the Role

We are looking for a Robotics Simulation Expert to own and evolve our simulation pipeline for robot learning and testing. This role sits at the intersection of physics simulation, 3D content pipelines, and machine learning building high-fidelity, physically accurate digital twins that power robot training, sim-to-real transfer, and synthetic data generation.

You’ll work across NVIDIA's Isaac ecosystem (Isaac Sim, and the Newton physics engine), 3D asset pipelines (USD, Blender), and modern scene-representation techniques (Gaussian Splatting) to build environments and assets that robots can learn and be validated in.

What You’ll Do

• Build, configure, and maintain robot simulation environments in NVIDIA Isaac Sim, including scene setup, sensor simulation, and robot/asset integration.
• Work with the Newton physics engine (NVIDIA's GPU-accelerated physics engine for Isaac) to configure and tune rigid-body, articulated, and contact-rich simulations for realistic robot dynamics.
• Develop pipelines to capture, process, and convert Gaussian Splat reconstructions of real-world scenes/objects into simulation-ready assets — including export, cleanup, retopology, and editing in Blender.
• Own the full asset pipeline: from raw capture Gaussian Splat reconstruction Blender editing/optimization export import into Isaac Sim publishing to Isaac Newton asset workflows.
• Understand and work directly with the internals of Universal Scene Description (USD) files — stages, layers, references/payloads, prims, schemas, variants — to author, debug, and optimize simulation scenes at a low level.
• Apply ML techniques to simulation workflows, e.g., domain randomization, synthetic data generation for perception models, sim-to-real gap reduction, and using ML for reconstruction/asset generation (including Gaussian Splatting-based methods).
• Support and where needed integrate with Gazebo for teams/projects still using it, ensuring interoperability or migration paths to Isaac Sim.
• Collaborate with robotics, perception, and ML teams to define simulation requirements, validate physical accuracy, and debug discrepancies between simulated and real-world behavior.
• Optimize simulation performance (physics step time, rendering, scene complexity) for large-scale/parallelized training runs.
• Document pipelines, tooling, and best practices for asset creation and simulation setup.

What We’re Looking For

Required:

• Master’s degree in Machine Learning, Robotics, Computer Science, or a related field.
• Hands-on experience with NVIDIA Isaac Sim for robotics simulation (scene building, robot import, sensor simulation, ROS/ROS2 bridges a plus).
• Working knowledge of the Newton physics engine (or equivalent GPU-accelerated physics engines) and general physics simulation concepts — rigid body dynamics, contacts/collisions, constraints, solvers.
• Practical ML experience, including exposure to Gaussian Splatting for 3D scene/object reconstruction, and comfort with optimization techniques (e.g., gradient-based optimization, loss design, training/inference workflows).
• Proficiency in Blender, including editing, cleaning, and optimizing 3D assets (meshes, materials, retopology) for real-time/simulation use, and exporting to simulation-compatible formats.
• Solid understanding of Universal Scene Description (USD) internals — stage/layer composition, prims, schemas, references and payloads — and the ability to author or troubleshoot USD directly.
• Experience taking assets through a full pipeline: capture/reconstruction edit export simulation import asset publishing.

Nice to Have:

• Familiarity with Gazebo for robotics simulation and experience migrating or bridging workflows between Gazebo and Isaac Sim.
• Experience with ROS/ROS2.
• Background in sim-to-real transfer, domain randomization, or synthetic data generation for perception/RL pipelines.
• Python/C++ proficiency for simulation tooling and pipeline automation.
• Familiarity with NVIDIA Omniverse and related USD-based content pipelines.