Perception Engineer

Applied Intuition • Ann Arbor, MI

11d • $125,000 - $175,000 • Onsite

About The Position

Applied Intuition Defense is hiring a hands-on, full-stack Perception Engineer to improve perception-model performance across maritime, land, air-to-surface, and surface-to-surface domains using electro-optical and infrared data. The team develops custom detector and classifier architectures, MLOps tools for iterating over millions of images and videos, labeling software, and in-house labeling processes. The work uses frontier AI and involves building systems and tools as needed.

The engineer will develop and deploy perception models, optimize model and inference performance on edge hardware, build foundation models to support assisted data labeling, create tools across the perception stack, and expand datasets using simulation and generative methods. The role involves cross-functional work with hardware, sensor, tracking, autonomy, and testing teams, including support for on-site Department of Defense testing and demonstrations. Employees are expected to work primarily from an Applied Intuition office five days a week; occasional remote work may be possible. Travel is generally 2–5 days every other month, estimated at 10–20%, as project needs require.

Requirements

• MS or PhD in Computer Engineering, Robotic Engineering, Computer Science, or an equivalent field, or 5+ years of relevant experience with simulation, machine learning, and ML infrastructure.
• Proficiency training ML models in PyTorch on multi-machine, multi-GPU systems.
• Experience optimizing and deploying machine-learning models to edge devices.
• Strong agentic coding experience and proven experience in Python and/or C++.
• Experience using modern AI-powered development tools, such as GitHub Copilot or Cursor, to create robust, well-tested systems.
• Ability to work with remote software development, handle and process large datasets, and learn new software and algorithms with little supervision.
• Must be a U.S. citizen.
• Must hold or be eligible to obtain and maintain a U.S. security clearance.
• Must be willing to travel as projects require, usually 2–5 days every other month (estimated 10–20%).

Nice To Haves

• PhD in Computer Vision, Machine Learning, or a related field, with strong academic contributions.
• 5+ years of experience building and deploying perception models on real-world hardware.
• Experience generating datasets using simulation engines such as Unreal Engine, NVIDIA Omniverse, or Blender, or using 3D asset creation tools.
• Experience deeply optimizing Transformer-based models for edge devices such as NVIDIA Jetson boards.
• Familiarity with Software-in-the-Loop (SIL) and Hardware-in-the-Loop (HIL), including integration and testing on common COTS hardware such as NVIDIA Jetson or Raspberry Pi 5.

Responsibilities

• Develop perception models that find and classify targets with an error rate below 1/100.
• Deploy perception models to Jetson Orin+ boards and optimize them for better-than-real-time performance.
• Build large foundation models for high-accuracy assisted data labeling.
• Build tools across the perception stack, including labeling tools and simulation.
• Scale datasets using state-of-the-art data-generation methods, including simulation, diffusion, and generative AI.
• Create and maintain highly optimized models and inference software to maximize frames per second on Jetson-like hardware.
• Collaborate with hardware, sensor, tracking, autonomy, and testing teams to support deployment at on-site DoD testing and demonstration events.

Benefits

• Equity in the form of options and/or restricted stock units may be included in the compensation package.
• Comprehensive health, dental, vision, life, and disability insurance coverage.
• 401(k) retirement benefits with employer match.
• Learning and wellness stipends.
• Paid time off.
• Relocation assistance may be available; interested candidates are encouraged to inquire.

What This Job Offers

• Job Type: Full-time