Senior Machine Learning Engineer (m/f/*)

About the role

About Peregrine Based in Berlin, we're leveraging the power of AI to transform cameras into intelligent devices, enhancing road safety and urban mobility on a global scale, while preserving privacy at all times. At Peregrine, diversity and international collaboration fuel innovation. We unite the brightest minds with strong academic backgrounds and rich industry experience in robotics and machine learning, spanning from the tech hubs of Silicon Valley to the engineering powerhouses of Europe. Our team boasts alumni from leading automotive giants like Bosch, Volkswagen, IAV, and TomTom and institutions such as the ETH in Zurich. We're on the hunt for brilliant, dynamic, and passionate individuals eager to tackle challenging problems and make a tangible impact. Tasks The Role As a Senior Machine Learning Engineer, you will own the design, training, and on-device deployment of the computer vision models at the heart of our product. You will work at the intersection of research and production, turning state-of-the-art vision techniques into reliable systems that run within strict latency and privacy constraints on resource-constrained edge hardware. Your key tasks will include: Design, train, and optimize deep learning models for object detection, semantic segmentation, pose estimation, and tracking. Port and deploy models to resource-constrained edge hardware, achieving single-digit millisecond latency without cloud dependencies. Build and maintain robust vision pipelines from data ingestion through training to production inference. Apply model compression techniques such as quantization, pruning, knowledge distillation, and neural architecture search to meet strict performance budgets. Develop synthetic data and domain adaptation pipelines to close the sim-to-real gap. Profile inference pipelines end-to-end to identify and eliminate bottlenecks on target silicon. Translate cutting-edge academic research into highly reliable, production-grade systems. Collaborate closely with hardware, product, and research colleagues to shape our privacy-by-design architecture. Requirements Your Profile Core Competencies: Computer Vision & Edge AI Edge AI & On-Device Inference: Expertise in porting, deploying, and optimizing complex deep learning models for local, resource-constrained hardware without cloud dependencies. Advanced Computer Vision: Deep knowledge of developing vision pipelines for object detection, semantic segmentation, pose estimation, and tracking. Hardware-Aware Architecture Design: Ability to custom-build network topologies tailored to specific sensors and strict latency budgets, rather than relying on off-the-shelf APIs. Synthetic Data & Domain Adaptation: Proven experience in building simulation pipelines for synthetic data generation and closing the “sim-to-real” gap using Domain Randomization and GANs. Technical Stack & Tools Languages: Advanced proficiency in C++ (for production-grade edge deployment) and Python (for training, research, and data analysis). AI/ML Frameworks: Extensive hands-on experience with PyTorch, and TensorFlow / Keras. Optimization & Deployment: Mastery of inference acceleration and model conversion using TensorRT, ONNX, and OpenVINO. Model Compression: Practical application of quantization, pruning, knowledge distillation, and neural architecture search. System Architecture & Compliance Privacy-by-Design Architecture: Designing robust, local AI systems that guarantee data sovereignty and strict GDPR compliance without cloud roundtrips. Performance Profiling: Deep-dive auditing of inference pipelines to identify bottlenecks and achieve single-digit millisecond latency on target silicon. Research, Strategy & Leadership Research-to-Production (R2P): The ability to translate complex, state-of-the-art academic research into highly reliable systems that work in the real world. AI Strategy & Assessment: Capability to conduct feasibilit

Skills