Computer Vision AI/ML Engineer

Leidos•Huntsville, AL

1d•Hybrid

About The Position

Serve as an AI/ML Engineer with a specialty in Computer Vision for the Contested Autonomy and Navigation Branch in Huntsville, AL. Our team’s mission is to: (1) develop and demonstrate new alternative navigation and vision processing algorithms to meet urgent military operational needs to operate in GPS-challenged environments, and (2) rapidly transition this technology to operational platforms to outpace the adversary threats currently being faced in the field. Projects involve small teams of engineers working together in spiral or agile development environments to meet aggressive customer schedules aimed at near-term field demonstrations of emerging capabilities. Our core competencies include sensor fusion, computer vision, HW/SW integration, and machine learning. Typical customers include Defense Advanced Research Projects Agency (DARPA), Army Combat Capabilities Development Command (CCDC) engineering centers, Office of Naval Research (ONR), and Air Force Research Laboratory (AFRL). We are looking for a candidate who is interested in applying classic and deep learning computer visions algorithms from initial concept development through successful prototype field demonstrations. Please note: This role is located in Huntsville, AL. There may be opportunity to telework up to 1 day/week. THE CHALLENGE: Develop perception algorithms using inputs from visible, infrared, and synthetic aperture radar sensors. Expected tasks include development of object detection, object tracking, image segmentation, optical flow, and crossview matching algorithms. Prototype initial algorithms in preferred programming language (i.e., Python, MATLAB, C++) Transition prototype algorithms for real-time edge execution on various System on Chips (CPU/GPU, CPU/TPU, etc.) Support field demonstrations of vision software on various manned and unmanned vehicles Present techniques, solutions, and results to DoD customers and Leidos management Develop proposals and concepts responding to customer requirements

Requirements

BS degree in a scientific field such as computer science, math, data science, physics, or engineering with 4+ years of prior relevant experience or Masters with 2+ years of prior relevant experience.
Experience in applying computer vision to solve complex image processing problems, including data curation, data labeling, model training, and model evaluation
Experience with state-of-the-art neural network architectures such as convolutional neural networks (CNNs) and vision transformers (ViTs) as well as specific high-performing open-source model implementations
4 years of experience with PyTorch (preferred), TensorFlow, or similar AI/ML development framework
3 years of experience with computer vision libraries such as OpenCV
3 years of experience in Python with additional experience in C++ and MATLAB
3 years of experience working in a Linux development environment
U.S. Citizen with the ability to obtain and maintain a Secret Security Clearance

Nice To Haves

Experience deploying to various edge processing devices including the NVIDIA Jetson family and other hardware acceleration platforms
Experience with Apache TVM or other similar frameworks for AI/ML acceleration for deployment to edge devices
Experience working with the Robot Operating System (ROS)
Experience also developing algorithms that process vision measurements such as Simultaneous Localization and Mapping (SLAM), visual odometry (VO), photogrammetry, and structure-from-motion (SfM)
Existing US security clearance (Secret or above)

Responsibilities

Develop perception algorithms using inputs from visible, infrared, and synthetic aperture radar sensors
Development of object detection, object tracking, image segmentation, optical flow, and crossview matching algorithms
Prototype initial algorithms in preferred programming language (i.e., Python, MATLAB, C++)
Transition prototype algorithms for real-time edge execution on various System on Chips (CPU/GPU, CPU/TPU, etc.)
Support field demonstrations of vision software on various manned and unmanned vehicles
Present techniques, solutions, and results to DoD customers and Leidos management
Develop proposals and concepts responding to customer requirements