Job Description

** This role is onsite in Fremont, CA with travel to Sacramento, CA several times per month.

Responsibilities

• Design and implement object detection, semantic/instance segmentation, and tracking pipelines for dynamic natural scenes

• Train and evaluate models using PyTorch or TensorFlow on curated datasets, including weakly-labeled or noisy real-world data

• Optimize models for edge deployment using TensorRT, pruning, quantization, and inference profiling

• Implement preprocessing, augmentation, and post-processing strategies for real-time use

• Evaluate and manage datasets as labeled by a separate labeling team

• Collaborate with systems and cloud engineers to deploy, test, and iterate models on Jetson devices

• Validate and analyze real-world performance, failure cases, and environmental variability

• Support integration of perception outputs with downstream autonomy logic and mission control

Required Qualifications:

• M.S. or Ph.D. in Computer Vision, Robotics, AI, or related field, or equivalent industry experience

• 3+ years of relevant experience designing and implementing object detection, semantic/instance segmentation, and tracking pipelines for dynamic natural scenes

Core

• Strong proficiency in Python for model development and deployment

• Experience with PyTorch or TensorFlow, including model customization and transfer learning

• Proven experience in object detection and segmentation (e.g., EfficientNet, Mask R-CNN)

• Deployment experience on Jetson platforms (Orin, Xavier, Nano) with constrained CPU/GPU budgets

• Familiarity with TensorRT, torchscript, or related deployment tools

• ROS2 experience including writing nodes, using CLI tools, and integrating with message types

• Experience developing and debugging software on embedded Linux

Infrastructure

• Familiarity with AWS or GCP for training, experimentation, and data management

• Experience with dataset versioning and model tracking (e.g., DVC, Weights & Biases, MLflow)

Preferred Qualifications:

• Experience with active learning, few-shot learning, or semi-supervised CV in natural settings

• Multi-camera fusion or 3D perception (stereo, depth, point cloud)

• Camera calibration, distortion correction, and temporal synchronization

• Understanding of outdoor vision challenges (lighting, weather, occlusion)

• Docker-based deployment of ROS2 systems to edge devices

• Contributions to open-source CV or robotics projects

• Prior work in autonomous outdoor robotics, agriculture, construction, or other rugged field environments

Nextracker offers a comprehensive benefits package. We provide health care coverage, dental and vision, 401(K) participation including company matching, company paid holidays with unlimited paid time off, generous discretionary company bonuses, life and disability protection and more. Employees in certain positions may be eligible for stock compensation. All plans are in accordance with relevant plan documents. For more information on Nextracker’s benefits please view our company website at www.nextracker.com Pay is based on market location and may vary based on factors including experience, skills, education and other job-related reasons. The annual salary range for this position is $180,000 - $220,000.