Lead DL Engineer for State Estimation & Vision

Shield; United States; On-site; USD 200,000 - 300,000; Full time

Job summary

Shield AI is seeking a qualified engineer to advance deep learning and 3D vision for autonomous navigation when GPS is unavailable. You will own model development, data preparation, and evaluation across diverse conditions, collaborating with state estimation and deployment teams.

You will work on pipelines for data ingestion, experiment tracking, and integration with VIO/terrain-relative navigation systems, with emphasis on reproducible training and robust performance.

Qualifications

• M.S. in Aerospace Engineering, Electrical Engineering, Robotics, Computer Science, or related field; 4+ years with M.S. or 2+ years with Ph.D.
• Hands-on experience designing, training, debugging, and evaluating models using PyTorch or equivalent, including architecture selection and loss design.
• Strong foundations in camera models, coordinate transformations, projective geometry, and multi-view geometry with practical experience in vision-based navigation, SfM, SLAM, or depth estimation.
• Strong Python skills and ability to turn research into working software; experience building data pipelines and reproducible training workflows.

Responsibilities

• Develop and evaluate models for feature detection, matching, depth estimation, and image-to-map localization.
• Combine learned representations with geometric methods to improve localization under varied conditions.
• Own data preparation, annotation needs, and dataset versioning; design robust training workflows and experiment tracking.
• Profile models for onboard compute, memory, and latency; collaborate with deployment teams on optimization.

Job description

Shield AI is seeking a qualified engineer to advance deep learning and 3D vision for autonomous navigation when GPS is unavailable. You will own model development, data preparation, and evaluation across diverse conditions, collaborating with state estimation and deployment teams.

You will work on pipelines for data ingestion, experiment tracking, and integration with VIO/terrain-relative navigation systems, with emphasis on reproducible training and robust performance.