Autonomy/Computer Vision Engineer

About the role

You will join a drone startup where autonomy is the differentiator: aircraft that see, understand, and navigate the world without a pilot's hands on the sticks. You build the perception and autonomy stack, from object detection running on edge compute to SLAM systems that keep the vehicle localized. Your algorithms decide what the drone does when the link drops and the GPS fades, which makes your work the ultimate backstop for every autonomous mission. You will work alongside the flight controls and navigation teams to ensure perception outputs integrate cleanly with autonomy logic, safety monitors, and fallback behaviors.

What you will achieve

• Ship onboard perception models for detection, tracking, and segmentation that run in real time on edge compute within strict power and thermal budgets.
• Build and validate visual navigation capabilities such as SLAM or visual-inertial odometry that maintain localization when GNSS is degraded or unavailable.
• Reduce autonomy failure rates by creating simulation and real-world test harnesses that stress perception across lighting, weather, and scene variation.
• Optimize model inference latency and memory footprint, hitting frame-rate targets on platforms like NVIDIA Jetson without sacrificing accuracy.
• Deliver autonomy features from prototype to flight-tested release, with documented safety cases and fallback behaviors for perception degradation.

What you will bring

Must-haves

• 2-5 years of experience in computer vision or robotics autonomy, with systems deployed on real hardware in real environments.
• Strong deep learning skills in PyTorch or TensorFlow, including training, evaluation, and deployment of detection or segmentation models.
• Working knowledge of SLAM, visual odometry, or multi-sensor fusion as applied to mobile robots or drones.
• Proficiency in C++ and Python, with experience optimizing code for embedded or edge platforms.
• Experience with camera calibration, sensor synchronization, and handling real sensor data with all its imperfections.
• Familiarity with ROS 2 and simulation tools such as Gazebo or Isaac Sim for autonomy development.
• Bachelor's or Master's degree in computer science, robotics, or a related field.

Nice-to-haves

• Experience with obstacle avoidance and path planning tightly integrated with perception outputs.
• Background in embedded GPU optimization using TensorRT, quantization, or pruning.
• Publications or meaningful open-source contributions in vision or SLAM.
• Experience with multi-camera or sensor-fusion perception architectures.