Machine Learning Engineer - Spatial AI

Do you want to dive in the fast-growing industry of drones and get a rewarding experience in a dynamic scale-up environment?

At Flyability, we believe that robots should be sent into hazardous places and dangerous environments instead of humans. To support our belief, we created Elios, the world’s first collision-tolerant flying robot that can safely enter, survey, and inspect confined spaces so that people don’t have to. With more than 150 employees and 1’500 customers,, Flyability is the market leader in the UAS indoor inspection industry.

Joining Flyability is not just taking on a new job; it is seizing the opportunity to improve the lives of millions of people and contribute to the future of robotics.

To complete our creative and dynamic team in Lausanne, we are seeking a:

Machine Learning Engineer - Spatial AI (100%)
Ideal starting date: as soon as possible

As a member of the Autonomy team, you will develop AI-based perception and spatial understanding for Flyability’s inspection drones. You will combine 2D images and 3D point clouds to build robust semantic representations of complex indoor environments.

You will work at the intersection of computer vision, machine learning, and robotics, connecting information across different views and sensing modalities. You will bridge research and engineering, turning ambitious AI ideas into robust capabilities for real-world robotic systems.

• Design and implement state-of-the‑art AI models and algorithms for 2D/3D scene understanding
• Combine camera and point‑cloud data into consistent semantic spatial maps, leveraging multi‑view and cross‑modal information
• Develop semantic understanding to support mapping, navigation, inspection planning, and natural‑language‑guided missions
• Prepare and analyze real‑world drone datasets
• Evaluate models using meaningful quantitative metrics, visualizations, and targeted experiments, and turn failure cases into model improvements
• Build reliable Python/PyTorch research and production pipelines, from experimentation to deployment
• Collaborate with mapping, sensors, and product teams to bring spatial AI capabilities into real‑world inspection systems

We are looking for an engineer who is excited by the challenge of making robots understand the physical world, and who can combine strong machine learning knowledge with solid engineering practices.

• 3+ years of experience developing machine learning or computer vision systems in an academic, industrial, or robotics environment
• Deep expertise in modern semantic segmentation, particularly in 2D imagery and/or 3D point clouds, including model architectures, training strategies, data preparation, and evaluation
• Proven experience developing production‑grade computer vision systems based on semantic segmentation, instance segmentation, or object detection, from data and training to evaluation and deployment
• Strong proficiency in Python and PyTorch, with the software engineering skills needed to turn research ideas into maintainable systems
• Strong understanding of the end‑to‑end ML lifecycle, including evaluating model performance on new environments, monitoring for degradation and emerging failure modes, and identifying whether improvements require better data, retraining, or model changes
• Experience with real‑world datasets, evaluation, and systematic error analysis, with a strong ability to turn failure cases into actionable model improvements
• An interest in 3D computer vision, multimodal perception, sensor fusion, or spatial AI, and in combining observations across cameras, point clouds, and multiple viewpoints
• The ability to move comfortably between research, prototyping, and production, and to turn promising ideas into robust robotic…