ML Data Engineer (m/f/d) - Sensor Data & Pipelines

Start free →

Or apply yourself on the official page →

About the role

What we offer • Work in an international, agile team creating the future of autonomous systems • Grow your career in a expanding and ambitious engineering team • Build innovative products using state-of-the-art technologies in AI, robotics, and autonomy  • Benefit from a steep learning curve and continuous development • Enjoy team events and a strong, collaborative culture Your mission This role owns the data foundation of our perception systems end-to-end — the layer that directly determines model performance in real-world environments. You'll set the technical direction for how we collect, curate, and continuously improve the datasets behind object detection, working as a senior technical partner to ML, perception, and robotics teams — turning raw, messy sensor data into reliable, production-grade systems at scale. You will take full ownership of the ML data lifecycle — from architecture decisions on ingestion and pipelines, through labeling strategy and QA, to driving continuous, metrics-informed dataset improvement — and will be expected to bring judgment and prior experience to how this is done, not just execute a defined process. What you'll do: • Architect and own scalable pipelines for ingesting, organizing, and preprocessing large volumes of time-series camera and multi-sensor data (RGB, IR, thermal, depth, IMU) • Drive the strategy behind our object detection datasets, ensuring quality, diversity, and statistical representativeness at scale • Design and operate active learning loops that connect model performance directly to data selection and improvement priorities • Own labeling workflows end-to-end — tooling decisions, QA methodology, consistency standards, and coordination of annotation efforts • Partner closely with AI Engineers to diagnose model weaknesses, bias, and drift, and translate findings into concrete dataset strategy • Plan and lead data collection campaigns (field recordings, drone/video capture) to close gaps with high-value real-world data • Build internal tools and dashboards that give the org visibility into dataset quality, distribution, and performance gaps Your profile • 5+ years of hands-on experience in Python and data processing frameworks (Pandas, NumPy, vectorized operations, multiprocessing) • Proven track record building and owning ETL/ELT pipelines for large-scale video and sensor datasets in production • Deep experience with data orchestration and lifecycle management for ML/computer vision workflows, including dataset versioning and reproducibility • Strong command of object detection pipelines (Detectron2, MMDetection, COCO format, bounding-box standards) • Demonstrated experience designing active learning, uncertainty sampling, or semi-supervised dataset workflows • Deep familiarity with data annotation platforms (CVAT, Label Studio) and building automated QA/consistency checks • Strong grasp of evaluation metrics for object detection (IoU, mAP, precision-recall curves, class-wise metrics) • Comfortable owning decisions around databases (SQL/NoSQL), file systems, and large-scale image, video, and sensor dataset management • Track record of working cross-functionally and influencing perception, deployment, robotics, and data infrastructure teams • Fluent in English; German and/or French are a plus Nice to have • Experience with cloud storage and MLOps tools (AWS S3, MinIO, ClearML, MLFlow, Weights & Biases). • Familiarity with ROS / robotics data formats (bag files, TF trees, sensor_msgs), Docker, or embedded ML workflows. • Prior work with robotics, drones, or multi-sensor perception systems, including IR, LiDAR, radar, or audio datasets. What else • Outside-the-box creativity with a blend of conceptual and systematic design thinking. • High intrinsic motiv

View the official posting →