Core contributor to AI-based Lidar and multimodal perception for L2-L4 autonomous driving systems, including BEV and multitask components in a 500+ developer international project. Improved 3D object detection under real-time constraints, addressing far-range, adverse weather and asynchronous multimodal fusion challenges, benchmarked architecture variants and training strategies, implemented automated data curation, built ML pipelines for data preparation, large-scale training, evaluation and model conversion, enabling real-time vehicle deployment and scalable offline auto-labelling across configurable multi-sensor setups. Developed tooling and pipelines to accelerate iteration cycles. Proposed and supervised an MSc thesis on LiDAR multitask perception resulting in a peer-reviewed publication, and started an internal AI reading group to foster knowledge sharing.

Research Fellow (Project: Bosch Factory of the Future: Smart Manufacturing) at University of Minho - BiRD Lab

February 1, 2021 - December 1, 2021

Developed real-time upper-body pose estimation from low-cost inertial sensors in a wearable system, combining sensor calibration, multi-sensor fusion and kinematic skeleton modeling, while exploring deep learning alternatives. Designed data acquisition and validation protocols, benchmarking system performance against a commercial motion capture solution. Deployed real-time algorithms within a ROS framework to enable closed-loop human–robot interaction.

Machine Learning Intern at Aquila Medical Innovation (Startup)

June 1, 2018 - September 1, 2018

Created NLP processing pipelines to extract and organize data from unstructured medical reports and applied deep learning models for X-ray image classification.