Leonardo Oliveira

Developed SLAM and localization software for indoor AMR fleets using sensor fusion between LiDAR and wheel odometry. Implemented robot behavior for intralogistics AMRs using behavior trees and developed ROS 2 perception components.

Main developer of an image denoising feature that significantly improved image quality while balancing cloud inference costs. Implemented camera auto-exposure algorithms and evaluated image quality across camera sensors and lenses.

Simulated autonomous driving scenarios using CARLA and Scenic. Trained and evaluated deep learning-based object detection algorithms, including YOLOv5, for research purposes.

Implemented EV charging algorithms to optimize energy costs by balancing external grid energy and locally generated solar panel energy. Developed low-level communication protocols in C++ and Node.js and integrated them into the internal software stack.

Worked on deep learning and natural language processing projects using Python. Built the backend of an API for a form-filler Chrome extension and developed data processing, ingestion, and monitoring pipelines using Apache Airflow.

Developed a vehicle emissions estimation model using machine learning, driving behavior data, and geospatial emissions data. Built ETL pipelines to automate statistical analyses.

Developed an algorithm to detect the absence of products on supermarket shelves. Set up and validated different deep learning-based techniques.

Grade: 1.5 (German scale, 1.0 best)

Master of Electrical and Computer Engineering with a final thesis on self-supervised depth completion with uncertainty and a research project on depth completion for LiDAR-visual-inertial SLAM.

Grade: 7.3/10

Bachelor of Electrical Engineering with a final thesis on seizure prediction using EEG data and machine learning, and a research project on CNN-based Poisson noise reduction in mammographic images.