Shield AI
Shield AI
Shield AI develops artificial intelligence systems for aerospace and defense applications. Its Hivemind platform supports autonomous mission capabilities, including intelligent drone operation, surveillance, and battlefield awareness. Founded in 2015, the company brings together expertise in AI, aerospace, and defense to build technologies intended to help protect service members and civilians.

Staff Deep Learning Engineer, State Estimation — Shield AI (Remote, United States)

Lead deep learning and 3D vision work for Shield AI’s autonomous defense systems. Develop localization, navigation, and Hivemind SDK capabilities for state estimation.

Description

  • Create and assess models for feature detection, matching, visual correspondence, depth, relative pose, and image-to-map localization.
  • Fuse learned visual features with geometric techniques to strengthen localization accuracy, resilience, and recovery.
  • Lead data preparation and supervision practices, from dataset curation and annotation specifications to labeling tools, automated quality controls, and coverage analysis.
  • Evaluate deep learning tools and establish reproducible training processes with experiment tracking, configuration control, and dataset and model versioning.
  • Build evaluations that account for variation in illumination, viewpoint, altitude, terrain, weather, and sensor properties.
  • Investigate model failures and direct improvements across data, supervision, modeling, and system integration.
  • Work with state estimation engineers to incorporate learned measurements and confidence estimates into VIO and terrain-relative navigation systems.
  • Measure models against onboard compute, memory, and latency limits, contributing to optimization and runtime verification.
  • Produce tested, documented components and interfaces for the Hivemind SDK.
  • Coordinate with software, systems, and flight-test teams.

Requirements

  • Master’s degree in aerospace engineering, electrical engineering, robotics, computer science, or a related discipline, plus at least 4 years of relevant professional experience; alternatively, 2 years with a Ph.D.
  • Practical experience designing, training, debugging, and evaluating models with PyTorch or a comparable framework.
  • Solid understanding of camera models, coordinate transformations, projective geometry, and multi-view geometry.
  • Applied experience in areas such as vision-based navigation, visual geolocation, Structure from Motion, SLAM, 3D reconstruction, or depth estimation.
  • Advanced Python skills and a record of producing maintainable, reusable software.
  • Experience advancing computer vision capabilities from problem definition and raw data through training, evaluation, and integration readiness.
  • Experience developing sensor-data pipelines for ingestion, cleaning, filtering, deduplication, and dataset version control.
  • Ability to choose development tools and create reproducible training workflows covering configuration management, experiment tracking, checkpointing, and GPU performance debugging.
  • Experience creating benchmarks, preventing data leakage, examining performance across operating conditions, and relating model metrics to downstream geometric or localization accuracy.
  • Ability to profile inference latency and memory consumption, document model interfaces and preprocessing, evaluate accuracy-versus-compute tradeoffs, and advise on export, precision, and runtime optimization.
  • Clear communicator who can explain assumptions, experimental results, and design tradeoffs while translating research into production software.
  • Offers are contingent on a cleared background check and may include a reference check.

Benefits

  • Bonus compensation.
  • Benefits package.
  • Equity participation.
  • Temporary employees receive a temporary benefits package after 60 days of employment.

Related Jobs