3 months ago
Clearwater, FL, USASenior
Responsibilities
- Build and maintain high-fidelity, physics-accurate simulation environments in NVIDIA Omniverse and Isaac Sim.
- Generate large-scale synthetic datasets with sensor and camera simulation, domain randomization, procedural scene variation, and automated annotations such as segmentation, depth, bounding boxes, and pose.
- Own dataset quality, versioning, and delivery.
- Design and run reinforcement-learning and imitation-learning pipelines using simulation-generated and synthetic data.
- Train and tune policies for control, planning, navigation, and manipulation, emphasizing robustness and sim-to-real performance.
- Define task curricula, reward functions, and evaluation benchmarks before policies reach hardware.
- Model sensors, actuators, and contact behavior while debugging simulation instability, non-physical behavior, and determinism issues.
- Drive simulation-to-real transfer through domain randomization, system identification, and physical-system validation.
- Build reusable tooling, APIs, and documentation for creating new environments and tasks.
- Integrate foundation models for reasoning, task decomposition, and human-in-the-loop learning.
Requirements
- 5+ years of experience in robotics, reinforcement learning, simulation, or applied machine learning.
- Degree in Robotics, Computer Science, or a related field, or equivalent hands-on experience.
- Hands-on experience training robotic agents in simulation, on physical systems, or both.
- Strong background in reinforcement learning, imitation learning, or learning-based control, including domain randomization and curriculum learning.
- Experience building simulation environments with NVIDIA Omniverse, Isaac Sim, Isaac Lab, or comparable GPU-accelerated simulation platforms.
- Direct experience generating synthetic training data, including sensor simulation, annotation pipelines, and large-scale dataset generation.
- Working knowledge of OpenUSD and asset conversion from formats such as URDF or MJCF into simulation pipelines.
- Production experience with at least one reinforcement-learning library: RSL-RL, RL-Games, skrl, or Stable-Baselines3.
- Strong Python and deep-learning-stack experience with PyTorch or JAX, including scaling training pipelines beyond a single workstation.
- Experience applying or integrating foundation models into robotics or decision-making workflows.
- Track record of moving learning systems from experimentation to deployment.
- Preferred experience with PhysX schemas and physics tuning; MuJoCo Playground, NVIDIA Warp, or Newton; Omniverse Replicator; world foundation models; vision-language-action models; ROS 2; real-time systems; physics engines; distributed or cloud-scale training; synthetic-data perception models; C++; and learned policies deployed on physical robots in production.
Benefits
- Remote or hybrid US-based work with periodic onsite time at the robotics facility.
- Occasional domestic and global travel.
- Flexible working hours aligned with experimentation and training cycles.
- Elective benefits tailored to the employee's country.
- Formal leadership and professional-development programs and on-demand courses.
- Financial, physical, and mental well-being support through seminars, events, and the global Life Empowerment Assistance Program.
- Inclusive communities, diversity and equity initiatives, peer-led activities, volunteering, and environmental and social programs.
- Onboarding opportunities to network with new coworkers within the first 30 days.
Categories
ML EngineeringRobotics
About TD SYNNEX
TD SYNNEX distributes IT hardware, software, and cloud services and aggregates solutions for technology vendors and resellers, including VARs and system integrators. It provides logistics, financing, cloud marketplaces, and professional services that help partners procure, bundle, and deploy technology for end customers. The public company (NYSE: SNX), headquartered in Fremont, California, was formed in 2021 through the merger of Synnex and Tech Data and operates globally.