João Gabriel Fernandes Zenóbio

- Developed a 3D hand tracking system for robot handover with Deep Learning and Computer Vision using MediaPipe and ROS2. - Developed a pose estimation system for human action detection with Deep Learning using YOLO and Computer Vision with multiple RGB-D cameras. - Created visualization and analysis tools using OpenCV and Open3D.

- Increased patient follow-up appointment rates using Generative AI (LLM). - Built automated ETL processes in Databricks workspace and using Terraform.

- Built a 2D and 3D dataset for unsafe behaviour in human-robot interaction using Computer Vision and RGB-D Cameras.

- Implemented new features in different Deep Learning Frameworks (Pytorch, Tensorflow, and JAX) and reviewed Pull Requests on Github.

- Automated the exploratory data analysis of animal behaviour images by creating custom annotated data, fine-tuning YOLO, compressing data with Autoeconders, and using Clustering Algorithms such as HDBSCAN*. - Optimized the data analysis time from days of work to seconds. - Increased the number of data correlated from a few to hundreds.