Actively hiring

Principal Perception Engineer

Start Date

Starts Immediately

CTC (ANNUAL)

₹ 4,00,000 - 5,00,000 ₹ 4,00,000 - 5,00,000 /year

Experience

1 year(s)

1 year(s)

Apply By

28 Oct' 26

Posted 1 week ago

Job

Be an early applicant

About the job

Key responsibilities:

Object detection and 6-DOF pose estimation
Design and maintain the primary perception pipeline (YOLO family / RT-DETR / FoundationPose or equivalent) for detecting and localising items in 6-DOF across cluttered shelf environments
Develop and benchmark instance-segmentation models that handle stacked, occluded, and unlabelled items under varying warehouse lighting
Implement and tune depth-completion and point-cloud filtering pipelines (passthrough, voxel-grid, radius outlier removal, RANSAC plane segmentation) on RGB-D camera output
Build and maintain a structured item dataset pipeline: data collection, annotation tooling (CVAT / Label Studio), augmentation strategy, and active-learning loops

Sensor fusion and state estimation
5. Fuse data from RGB-D cameras, 2D LiDAR, and a 6-axis IMU using an Extended Kalman Filter (EKF) or factor-graph frameworks (GTSAM / iSAM2)
6. Manage temporal synchronisation across sensors with different update rates, using hardware-triggered or software-timestamp-based sync
7. Build covariance-aware fusion pipelines that degrade gracefully on sensor drop-out and pass meaningful uncertainty estimates to the motion planner
SLAM and localisation
8. Develop and tune the SLAM stack (SLAM Toolbox, Cartographer, or LIO-SAM) for a compact mobile base in narrow aisles with dynamic obstacles (people, forklifts)
9. Build map-management tooling: incremental map updates, multi-floor map stitching, and map-version control integrated with the ROS 2 / Nav2 stack
Edge deployment and grasp interface
10. Optimise perception models for NVIDIA Jetson-class edge hardware: TensorRT quantisation (INT8/FP16), ONNX export pipelines, model pruning, and layer fusion
11. Profile and maintain an end-to-end perception latency budget for real-time operation using NVIDIA Nsight Systems and ROS 2 tracing
12. Publish standardised pick-point and approach-vector messages to the motion planner (MoveIt 2 / custom Cartesian IK) with full uncertainty representation
Testing and MLOps
13. Build ROS 2 bag replay pipelines and Gazebo sensor plugins for regression testing without physical hardware
14. Manage model versioning, experiment tracking (MLflow / Weights & Biases), and a model registry integrated with CI/CD

Please Note-: Salary negotiable for the right professionals (Experienced Engineers)

Skill(s) required

Artificial Intelligence Markup Language (AIML) Autodesk Fusion 360 Computer Vision C Programming Docker Linux Machine Learning OpenCV PyTorch Robot Operating System (ROS) SolidWorks TensorFlow

Earn certifications in these skills

Who can apply

Only those candidates can apply who:

1. have minimum 1 years of experience

Other requirements

Mandatory: hands-on experience building physical robots, meaning real hardware you designed, built, and made work. Simulation-only or coursework-only projects will not be considered.

Final-year and pre-final-year https://jobeax.com/link/2DENLGgQM7whDhaE students and https://jobeax.com/link/s1WvHFV9ITSwCkyS students are preferred. PhD students in any year are welcome to apply.

Exceptional proven builders with a credible track record of building robots will be considered regardless of academic qualification.

Experienced engineers are also invited to apply. The prerequisite is physical robots built and delivered on commercial projects. Salary for experienced engineers is negotiable.

On-site in Bangalore.

Salary

Probation:

Duration:

Salary during probation:

After probation:

Annual CTC: ₹ 4,00,000 - 5,00,000 /year

Perks

Informal dress code 5 days a week

Number of openings

1

About Ometaura AI

We develop advanced robotic systems in real-world environments. Work includes designing mobile robotic platforms, including single-arm and dual-arm robots capable of autonomous and semi-autonomous operation. We build robotic manipulation systems for picking, placing, and handling objects in structured and semi-structured environments, along with perception and navigation stacks combining vision, sensing, mapping, localization, and motion planning. In parallel, we are building AI-driven applications designed to improve everyday life and optimize business and industry through powerful yet intuitive technologies. The founding team previously built and operated a robotics company incorporated in the United States.

Activity on Internshala

Hiring since December 2025

8 opportunities posted

Additional Questions

Sign up to continue

Candidate sign up

Sign up/ Login with Google

Sign up with Email

OR

By continuing as a candidate, you agree to our T&C .

Already registered? Login