Tomas Alesandro Poeta

Initially, focused on enhancing internal tools for data preparation and processing. I contributed to the development of a custom image labeling tool adding features and resolving existing bugs. Additionally, I optimized a dataset reviewer tool and rebuilt a dataset generator tool, leveraging the company's models to produce new images. Currently, my responsibilities include training and testing deep learning models such as YOLO, Faster R-CNN, and DETR, as well as integrating them into the main pipelines.

Co-leading the creation and implementation of a scalable Manual QA testing suite for the largest Europe cinema operator company. Initially focused on system analysis, test case design, and manual execution for cinema-specific modules and web. Advanced to coordinating the operational tasks of the Spanish/Portuguese UAT team on the website, having to assign tasks, supporting team members during testing, and report status/results of executions.

Labeled a medical dataset of medical images using CVAT.ai and by means of Python implemented AI models like U-Net for semantic segmentation. This work established the baseline for future research and formed the foundation of my Master's thesis.

Led the creation of an automated testing suite from inception, focusing on E2E, API, and performance tests. This included designing the suite's architecture, developing and maintaining the test cases. Also, I built a Jenkins pipeline for integration. I also managed manual testing and assumed QA leadership at the end. The initiative replaced an outdated testing suite, enhanced software quality across frontend, API, and performance.

Development of a software for a digital school using Python and Selenium performing web scraping to automatically assess student tasks with the data coming from a virtual platform.

As a consultant Java developer, I worked on migrating the architecture for a critical invoice processing area for a major company in Spain's tourism industry. Using Spring Boot and Kafka, we developed microservices to manage the entire process, from provider-generated invoices to their integration into the company's ERP system. This involved standardizing different invoice formats, loading them into a BI database, and ensuring their transfer to the ERP system.

Created alarms by means of Java to correlate information, collecting activity logs from various sources. After gaining expertise, I was assigned to develop plugins for Graylog to handle specific data sources beyond its native capabilities. This preprocessed information from sources, standardized data, correlated information using Java, and automated manual data entry tasks. Finally, deploying these plugins, reducing repetitive work, and creating a standardised method for processing specific log sources.

Initially focused on system maintenance and gradually took on developing new functionalities. My role involved working with Java, frontend development, and database management. My responsibilities included bug fixing, analysis and feature development, and various database tasks.

Member in the creation of a framework that would allow the creation of Augmented Reality applications. Performing Computer Vision tasks, specifically Pose Estimation and researching mechanisms to send the information obtained from such process to a Graphic Engine. Results: - The structure of the framework was designed. - Prototype applications were created to determine what functionality would be needed for the framework. - Two academic articles were written as a progress report.

Part of a small startup where we took on small to medium-sized projects. Being responsible for developing the frontend for websites, utilizing HTML5, CSS (SASS), and jQuery.

During this program, I gained a robust theoretical foundation and hands-on experience in Image Processing and Analysis, as well as for Visual Recognition and Modeling. I also developed expertise in Machine Learning, applying both foundational and advanced techniques specifically tailored to the field of Computer Vision.