Working Student - Software Engineering (m/f/d)
About this role
As a working student in Software Engineering, you support the annotation and quality assurance of training data as well as the structured management of our datasets. You also contribute to the definition and evaluation of KPIs for assessing our models and assist in conducting end-to-end tests of the entire system pipeline. In addition, you research new approaches in the field of computer vision and compare them with existing solutions. Furthermore, you help to further develop our testing infrastructure and evaluation pipelines to ensure continuous quality assurance.
What is your Day to Day Mission:
Data Annotation & Dataset Management – Annotate and quality-check training data for object detection and end-to-end planner models, including bounding boxes, segmentation masks, and trajectory labels; maintain and version datasets to ensure consistency and traceability across model iterations
KPI Definition & Model Evaluation – Design, implement, and continuously improve KPIs to monitor and benchmark the performance of ML models deployed on UAVs and UGVs, covering metrics such as detection auracy, latency, precision/recall, and robustness under real-world conditions
End-to-End System Testing – Execute and automate end-to-end integration tests of the full perception and planning pipeline on target hardware (Nvidia Jetson), identifying bottlenecks and regressions across the entire system stack
Research & Model Benchmarking – Independently investigate state-of-the-art publications and open-source models in computer vision, object detection, and autonomous systems; systematically compare novel approaches against our current solutions and document findings in a structured way
Tooling & Test Infrastructure – Con
Responsibilities
- Data Annotation & Dataset Management – Annotate and quality-check training data for object detection and end-to-end planner models, including bounding boxes, segmentation masks, and trajectory labels; maintain and version datasets to ensure consistency and traceability across model iterations
- KPI Definition & Model Evaluation – Design, implement, and continuously improve KPIs to monitor and benchmark the performance of ML models deployed on UAVs and UGVs, covering metrics such as detection auracy, latency, precision/recall, and robustness under real-world conditions
- End-to-End System Testing – Execute and automate end-to-end integration tests of the full perception and planning pipeline on target hardware (Nvidia Jetson), identifying bottlenecks and regressions across the entire system stack
Requirements
- Research & Model Benchmarking – Independently investigate state-of-the-art publications and open-source models in computer vision, object detection, and autonomous systems; systematically compare novel approaches against our current solutions and document findings in a structured way
- Tooling & Test Infrastructure – Con