We are IronBrain, a revolutionary JV on a mission to transform the Defense Industry with cutting-edge Autonomous AI technology. We are backed by the expertise of two leading companies: a leader in aerospace and defense solutions, and Cortica, a global leader in Autonomous AI.

Join our Core Team

As an AI & CV Algorithm Developer, you'll play a pivotal role in shaping the future of IronBrain. You'll be part of a team responsible for developing and implementing state-of-the-art AI based solutions for computer vision applications.

What We Do

Our current focus is on aerial computer vision, leveraging Synthetic Aperture Radar (SAR) from satellites and UAVs, as well as camera data from drones, UAVs & border cameras. We're building the foundation for the future, with plans to expand into radars, UGVs, robotics, and more.

What You'll Do

• Develop cutting-edge CV AI models for perception tasks : object detection, classification, to representation encoders, zero-shot & few-shot tasks.
• Solve data-scarcity challenges using advanced AI methods such as SSL pre-training, PEFT domain-shift, etc.
• Implement supplementary computer vision algorithms like homographies, tracking, and logic functions
• Create benchmarks to measure algorithm performance
• Implement peripheral software tools & infrastructure
• Be an integral part of our founding core team, shaping the company's technological direction

Requirements

• 4+ years of experience as an AI Algorithm SW developer in the industry
• 3+ years of experience in relevant AI algorithm domains (SSL utilisations, vision encoders, object detection / classification, etc.)
• Proven expertise in AI-based computer vision.
• Deep understanding of classical computer vision techniques and their applications
• Proven ability to design, implement, and evaluate complex AI systems
• Frameworks : Linux, Python, PyTorch, OpenCV, Docker

Advantages

• Experience in aerial vision (drones, UAVs, satellites, EO, SAR, IR)
• Experience in representation learning (embeddings/feature vectors)
• Experience with few-shot learning and self-supervised learning