Staff Deep Learning Engineer, State Estimation (R5785)

The role

You will build deep learning models that help autonomous systems understand their motion and find their location when GPS is unavailable or unreliable. You will own the model pipeline, from data and tools through training, evaluation, and deployment recommendations, on the Hivemind SDK State Estimation & Vision team.

What you will do

• Develop and evaluate models for feature matching, depth estimation, relative pose estimation and image-to-map localization.
• Combine learned visual models with geometric methods to improve localization accuracy and robustness.
• Own dataset curation, annotation requirements, labeling tools and quality checks.
• Build reproducible training workflows with experiment tracking and dataset and model versioning.
• Design evaluations across changes in lighting, viewpoint, altitude, terrain, weather and sensors.
• Analyze failures and run controlled experiments to prioritize improvements.
• Work with state estimation engineers to integrate learned measurements into VIO and terrain-relative navigation.
• Profile models against onboard compute, memory and latency limits, and work on optimization with deployment engineers.
• Deliver tested, documented components and interfaces for the Hivemind SDK.

Required qualifications

• M.S. in Aerospace Engineering, Electrical Engineering, Robotics, Computer Science or a related field, with 4+ years of related professional experience; or a Ph.D. with 2+ years.
• Hands-on experience designing, training and evaluating models in PyTorch or an equivalent framework.
• Strong foundations in camera models, coordinate transformations, projective geometry and multi-view geometry.
• Practical experience in one or more of: vision-based navigation, visual geolocation, Structure from Motion, SLAM, 3D reconstruction, depth estimation or similar fields.
• Strong Python skills and experience writing maintainable, reusable software.
• Experience building pipelines for sensor data ingestion, cleaning, filtering, deduplication and dataset versioning.
• Ability to build reproducible training workflows, including configuration management, experiment tracking and GPU performance troubleshooting.
• Experience designing benchmarks, preventing data leakage and connecting model metrics to localization accuracy.
• Ability to profile inference latency and memory and advise on export, precision and runtime optimization.
• Ability to communicate findings and design tradeoffs clearly.

Nice to have

• Experience with aerial imagery, geospatial data, elevation maps or cross-viewpoint and cross-sensor matching.
• Experience with model export, quantization, TensorRT, ONNX or embedded compute deployment.
• Experience validating perception or robotics systems on physical platforms.
• Publications, open-source contributions or production computer vision systems.
• Familiarity with correlation volumes, learned scene priors, or diffusion and flow matching for vision.
• Experience in the aerospace and/or defense industry.

Benefits

• Equity for full-time regular employees.
• Bonus for full-time regular employees.
• Benefits package for full-time regular employees.
• Temporary benefits package for temporary employees, applicable after 60 days of employment.

Before you apply

• Offers are contingent on a cleared background check and possible reference check.
• Military fellows and part-time employees are not eligible for benefits.
• Salary is influenced by skill set, experience, licenses, certifications and work location.

About the company

Shield AI is a venture-backed defense-tech company that builds intelligent systems to protect service members and civilians. Its products include Hivemind autonomy software, V-BAT and X-BAT aircraft, and Aechelon simulation technologies.