Perception Engineer - Multimodal & Embodied AI

About the role

Shield AI's Perception team is looking for an engineer who enjoys turning AI research into capabilities for autonomous aircraft and vehicles operating in the real world. Working alongside researchers and autonomy engineers, you will evaluate approaches, build prototypes and bring successful ideas into the Hivemind product, combining learned methods with established perception techniques.

Key responsibilities

• Evaluate emerging models and methods against representative sensor data and autonomy needs.
• Build prototypes and experiments that reveal strengths, limitations and deployment opportunities.
• Combine learned and classical approaches to improve detection, tracking, geometric vision or sensor fusion.
• Optimize and integrate capabilities for edge deployment, contributing Python and C++ software with support from experienced product engineers.
• Help deliver tested, documented capabilities and stay involved as other teams adopt them.

Required profile

• PhD in computer vision, machine learning, robotics or a related field, or an equivalent record of advanced research and hands-on engineering.
• Research depth in areas such as multimodal vision, visual foundation models, video understanding, open-vocabulary perception or embodied learning.
• A thoughtful approach to experiments, evaluation data and understanding why models fail.
• Experience building substantial research software, an interest in production-quality C++ and a collaborative approach.

Required skills

• Strong Python skills and experience with PyTorch or a comparable ML framework.
• Preferred: experience bringing a research prototype into sustained use, robotics, real sensors, temporal data or multi-sensor perception.
• Preferred: familiarity with vision-language or vision-language-action models, and C++, edge inference or model optimization with ONNX, TensorRT or CUDA.