Machine Learning Engineer - Spatial AI

Flyability, a Swiss scale‑up that builds collision‑tolerant inspection drones, is looking for a Machine Learning Engineer, Spatial AI to join its Autonomy team in Lausanne. The company's flagship robot, Elios, enables safe surveying of hazardous indoor spaces, serving more than 1 500 customers worldwide. In this role you will create AI‑driven perception and spatial understanding for the drones. Your daily work will involve designing and implementing state‑of‑the‑art models that fuse 2D camera images with 3D point‑cloud data to generate robust semantic maps of complex indoor environments. You will prepare real‑world datasets, run systematic experiments, evaluate performance with quantitative metrics and visualisations, and turn failure cases into concrete model improvements. The engineered pipelines, built in Python and PyTorch, will move from research prototypes to production‑ready components that support mapping, navigation, inspection planning and natural‑language‑guided missions, in close collaboration with mapping, sensor and product teams. The ideal candidate brings at least three years of experience developing machine‑learning or computer‑vision systems in an academic, industrial or robotics setting. You have deep expertise in modern semantic segmentation for both 2D imagery and 3D point clouds, and a proven track record of delivering production‑grade computer‑vision solutions that cover the full lifecycle from data preparation to deployment. Strong Python and PyTorch skills, a solid grasp of end‑to‑end ML processes, including monitoring, degradation detection and iterative improvement, and hands‑on experience with real‑world datasets and error analysis are essential. An interest in 3D vision, multimodal perception and sensor fusion, together with the ability to translate research ideas into reliable robotic capabilities, is required. Fluency in English is mandatory; French is a plus. Additional assets include experience with multi‑view geometry, 3D reconstruction, SLAM or semantic mapping, as well as deploying AI models on resource‑constrained robotic or embedded platforms. Backgrounds in robotics or autonomous platforms, a strong interest in linking semantic maps to natural‑language interaction, and familiarity with LLM fine‑tuning, retrieval‑augmented generation and multimodal evaluation are considered nice‑to‑have. Flyability offers 25 vacation days plus public holidays, flexible working hours and the option to work remotely up to two days per week. Employees benefit from comprehensive accident insurance, gym‑membership discounts, and access to the Swibeco rewards platform. Team events such as ski weekends, summer barbecues and after‑work gatherings foster a collaborative culture where every voice is heard. What the role asks for: - 3+ years ML or computer vision development experience - Deep expertise in modern semantic segmentation (2D/3D) - Production‑grade CV systems (segmentation, detection) deployment experience - Strong Python and PyTorch proficiency - End‑to‑end ML lifecycle knowledge (evaluation, monitoring, retraining) - Real‑world dataset handling and systematic error analysis - Interest in 3D vision, multimodal perception, sensor fusion - Ability to move from research to production robotics - English proficiency; French a plus - Experience with multi‑view geometry, 3D reconstruction, SLAM (nice to have) - Deploying AI models on resource‑constrained robotic platforms (nice to have) - Working with robotics or autonomous systems (nice to have) - Interest in semantic maps linked to natural‑language interaction (nice to have) - LLM fine‑tuning, RAG, evaluation for multimodal AI (nice to have)