Resume

Robotics Engineer

• ▸Developing the real-time perception stack for autonomous wheel loaders in unstructured off-road environments, taking research prototypes to on-vehicle production systems.
• ▸Co-authored sensVLA, a spatially-grounded Vision-Language-Action model for autonomous heavy machinery — accepted at the ICRA 2026 Workshop on VLA Pipelines for Real Robots.
• ▸Own end-to-end deployment of deep learning perception models on embedded GPUs, including quantisation, TensorRT engine builds, and CUDA-based pre/postprocessing.

Perception Engineer – HPC

• ▸LiDAR 3D Object Detection. Shipped a production 3D detection model to the test vehicle by engineering a TensorRT inference engine and custom CUDA pre/postprocessing kernels — reaching 24 FPS on the embedded GPU and meeting real-time constraints.
• ▸LiDAR Lane Segmentation. Built the end-to-end inference pipeline with a custom CUDA NMS kernel for lane outputs, achieving ~20 FPS on the edge device and unblocking on-road deployment.
• ▸Data & Annotation Tooling. Designed the ODD data-collection pipeline and built an in-house 3D LiDAR annotation tool, then led the in-house labelling team — scaling training-data throughput for downstream perception models.

Master's Thesis Student

• ▸Radar Moving-Object Segmentation. Designed a novel temporal transformer network for sparse radar point clouds; trained on the in-house Porsche Radar Dataset and RadarScenes.
• ▸Achieved state-of-the-art results with a 12% improvement over the prior baseline; work published in IEEE Transactions on Robotics (T-RO) under Prof. Dr. Cyrill Stachniss (IPB Lab, University of Bonn).

Graduate Student Assistant (HiWi)

• ▸Researched LiDAR intensity calibration methods to improve ego-vehicle localisation accuracy, and implemented a non-learning-based approach for dynamic-object removal from LiDAR scans.

Computer Vision Intern

• ▸Developed an identity tracking and motion prediction stack for a robotic camera slider and deployed the model on an NVIDIA edge device at 60 FPS.

Deep Learning Intern

• ▸Built a CNN-based screw-density classifier for industrial camera images and integrated it into the operator UI via a WebSocket service — enabling full automation of the machine.

M.Sc. in Mobile Sensing and Robotics — Grade 1.4

Focus: SLAM, 3D Object Detection, Point Cloud Analysis, Bundle Adjustment, CUDA, Deep Learning for Vision.

B.Tech. in Electronics and Communications Engineering — Grade 1.7