Research Engineer, Foundation Model (Multimodal Fusion & Distillation)

Laelaps AI is seeking a Research Engineer, Foundation Model (Multimodal Fusion & Distillation) to develop advanced AI models for physical security. The company specializes in creating intelligent software that powers autonomous systems, enabling robots and sensors to handle critical and dangerous tasks. Their mission is to enhance security and resilience in the real world through innovative AI solutions. The main responsibilities of the role include building and adapting open vision and point-cloud foundation models to various sensors such as RGB, IR, thermal, and lidar. The engineer will also develop a fusion layer that aligns all sensors in time and space, fine-tune a vision-language model to reason over fused sensor tokens, and create a fast model that processes every frame efficiently. Additionally, the role involves setting up the training, evaluation, and deployment processes for these models, ensuring they are tested and validated on real security sites. To qualify for this position, candidates must have at least four years of experience in training and fine-tuning large vision or multimodal models, or a PhD plus industry experience. Strong proficiency in PyTorch, particularly with multi-GPU and multi-node training, is essential. Experience in fine-tuning vision-language models, knowledge distillation, and working with non-RGB sensors like thermal, infrared, lidar, or radar is also required. Candidates should demonstrate a strong evaluation discipline and solid software engineering practices. Nice-to-have skills include experience with point-cloud or 3D foundation models, self-supervised and latent-prediction methods, edge deployment with TensorRT, and robotics or field data collection. Publications at top ML, vision, or robotics venues are also valued. The role is based in Zurich, offering a collaborative environment with a strong founding team, including PhD-level co-founders in AI, Robotics, and Physics. The company provides a competitive equity/salary package and a culture that is both serious about building and fun. What the role asks for: - 4+ years training and fine-tuning large vision or multimodal models, or a PhD plus industry experience. - Strong PyTorch, comfortable with multi-GPU and multi-node training. - Hands-on fine-tuning of vision-language models. - Experience with knowledge distillation or model compression for deployment. - Experience with video or non-RGB sensors: thermal, infrared, lidar or radar. - Evaluation discipline and solid software engineering hygiene. - Nice to have: Point-cloud or 3D foundation models. - Nice to have: Self-supervised and latent-prediction methods.