Open position
Senior Software Engineer (Perception)
We are seeking a Senior Software Engineer to build the vision pipelines behind AMP manipulation solutions. The near-term targets are depalletization and bin picking.
The pipelines run on production hardware in facility lighting against mixed loads, and they are measured on cycle reliability over a shift.
You will work under the AI Lead alongside the manipulation team, and own perception from sensor data through pose output.
What you’ll do
- Build and own perception pipelines for depalletization and bin picking, from raw sensor data to object and grasp pose.
- Implement 3D perception: point cloud processing, segmentation, pose estimation, and sensor fusion across RGB-D and LiDAR.
- Train, evaluate, and deploy models under the compute and latency constraints of the target hardware.
- Build the evaluation methodology that predicts field reliability, including failure case analysis.
- Handle the difficult cases directly: reflective and transparent surfaces, mixed SKU pallets, occlusion, and shifted loads.
- Integrate perception output with motion planning and grasp execution, and define the interface contract between them.
- Work with the Data Collection Lead to specify the data the pipeline actually needs.
- Instrument perception in the field so degradation is detected before the customer reports it.
Job requirements
Experience
- 4 to 8 or more years of professional engineering experience, with at least 2 years in applied computer vision or 3D perception.
- Experience deploying a perception system on physical hardware in an uncontrolled environment.
- Experience with perception for manipulation, including grasp-relevant output.
Education
- Master's degree or higher in Robotics, Computer Science, Electrical Engineering, or a closely related technical field.
- Equivalent industry experience with demonstrated perception depth may be considered.
Core technical background
- Strong background in 3D computer vision: point cloud processing, registration, segmentation, and 6DoF pose estimation.
- Experience with deep learning for detection, segmentation, and pose, plus the classical methods that often work better.
- Experience with RGB-D and LiDAR sensors, including calibration, synchronization, and their failure modes.
- Experience with model optimization and deployment on edge compute, including quantization and TensorRT or equivalent.
- Proficiency in Python and C++, with PyTorch and Open3D, PCL, or OpenCV.
- Hands-on experience with ROS or ROS 2 perception integration.
Systems mindset
- Ability to reason across the whole pipeline: sensor placement, lighting, calibration, and downstream motion planning.
- Treats latency and determinism as requirements.
- Diagnoses field failures from logged data.
Collaboration
- Able to work effectively with remote teams across multiple time zones.
- Comfortable working closely with manipulation and controls engineers.
Nice to have
- Experience with depalletization, bin picking, or singulation in production.
- Experience with grasp synthesis and grasp quality evaluation.
- Experience with vision language models applied to manipulation.
- Experience with active perception or next-best-view planning.
- Prior open-source contributions to perception tooling.