Talent.com
Senior Robotics Ml Engineer (Reinforcement Learning, Ackermann Vehicles)

Senior Robotics Ml Engineer (Reinforcement Learning, Ackermann Vehicles)

For a Zoho group of companyChennai, Republic Of India, IN
3 days ago
Job description

About the Job

A Chennai based Zoho group of company is seeking an experienced engineer to design, train, and deploy reinforcement learning (RL) policies for an Ackermann-steered robotic vehicle. This role involves full stack ownership from data pipelines and simulation to on-vehicle inference spanning perception, planning / control, and MLOps.

You should be skilled in PyTorch (primary), with production experience in TensorFlow / Keras, and possess strong data engineering capabilities.

What You’ll Do

Reinforcement Learning

  • Design and implement RL / IL algorithms for lateral and longitudinal control (e.G., SAC, TD3, PPO, IQL / CQL, BC / DAgger).
  • Build reward functions and constraints that respect Ackermann kinematics, curvature limits, tire slip, and comfort.
  • Contribute to safe and constrained RL frameworks with uncertainty estimation and fallback control strategies (PID / MPC).

Perception and Vision

  • Develop perception models for occupancy, drivable space, and obstacle detection / tracking.
  • Maintain computer vision pipelines for applications such as orchards, dairy farms, and off-road environments.
  • Fuse multi sensor inputs (camera, IMU, encoders, and optionally LiDAR) for robust state estimation.

    • Planning and Control Integration

  • Implement trajectory tracking, safety envelopes, and collision checking.
  • Calibrate steering angle to curvature (Ackermann geometry) and validate with off-road telemetry.
  • Collaborate on hybrid IL / RL control or MPC-based integration strategies

    • Data engineering and MLOps

  • Build data pipelines from vehicle logs to train / evaluation datasets with automated quality checks.
  • Design and manage feature stores, dataset versioning, and automated labeling loops.
  • Establish reproducible model training, experiment tracking, and CI / CD workflows for ML models

    • Simulation and testing

  • Demonstrate closed-loop on-vehicle driving at low to moderate speeds with defined safety gates and KPIs (tracking error, intervention rate, comfort).
  • Author scenarios and evaluators (closed‑loop tests, Monte Carlo, rare‑event mining).

    • On‑vehicle deployment

  • Deploy and optimize networks on embedded platforms (Jetson Orin / Xavier, x86 GPU) using TensorRT / CUDA with real-time scheduling and profiling..

    • Collaboration

  • Work cross‑functionally with controls, firmware, and test teams;

    • mentor junior engineers.

    Minimum Qualifications

  • 6+ years of industry experience in ML / robotics systems (or 3+ with a relevant MS / PhD), including 2+ yeas in RL for control / robotics.
  • Strong proficiency in Python and at least one systems language (C preferred).
  • Deep expertise in PyTorch with production exposure to TensorFlow / Keras.
  • Solid foundations in :

    • ■ RL / IL (value / policy gradients, offline RL, dataset curation, covariate shift handling).

    ■ Control and estimation (Ackermann kinematics / dynamics, PID / MPC, EKF / UKF).

  • ■ Computer vision (detection, segmentation, tracking;

    • BEV / occupancy).

  • Strong data engineering background—ETL, large-scale dataset handling, Docker / Linux, and GPU / cloud workflows.
  • Proven ROS / ROS2 development experience with real-world sim-to-real deployments on mobile robots or vehicles.

    • Preferred Qualifications

  • Experience with safe or constrained RL, uncertainty modeling, or risk-aware planning.
  • Background in offline RL (CQL / IQL / AWR) and hybrid IL / RL training curricula.
  • Familiarity with mapping / localization (HD / vector maps, lane graphs).
  • Experience in automotive / robotics safety (SOTIF ISO 21448, ISO 26262) and test track operations.
  • Experience with nav2, CAN, and embedded interfaces.

    • Key Outcomes (First 90–180 Days)

    Establish a data pipeline from vehicle logs to curated train / eval datasets with automated quality checks.

  • Deliver a baseline IL policy (lane following, obstacle avoidance) in simulation and progress to RL fine-tuning with safety constraints.
  • Demonstrate closed-loop on-vehicle driving at low to moderate speeds with defined safety gates and KPIs (tracking error, intervention rate, comfort).
  • Set up CI / CD workflows and reproducible benchmarking across simulation and track runs.

    • Stack You’ll Use

    ML : PyTorch (primary), TensorFlow / Keras, ONNX / TensorRT, CUDA Vision : OpenCV, torchvision, Detectron / YOLO, BEV / occupancy frameworks Robotics : ROS2, nav2, C17 / 20 Data & MLOps : Python, Pandas / Numpy, Arrow / Parquet, DVC / MLflow / W&B, Docker, Git, CI Simulation : CARLA, Isaac Sim, Gazebo, AirSim

    Create a job alert for this search

    Senior Ml Engineer • Chennai, Republic Of India, IN