
Contributed to the development of an education-focused Artificial Intelligence platform for the Portuguese public administration. Conducted rigorous testing and benchmarking of Large Language Models (LLMs) to evaluate their proficiency and performance in European Portuguese (PT-PT).

Led the end-to-end development of an AI-powered legal platform, transforming a basic prototype into a fully functional system. Engineered the project architecture, implemented testing frameworks, and systematically evaluated and integrated various AI models to advance the platform's capabilities.

Built a smart tourism mobile application in Flutter, integrating geofencing for real-time location tracking and gamification mechanics. Leveraged generative AI models (Gemini) to create contextual audio guides and a responsive virtual assistant for point-of-interest interactions.

Engineered a multi-agent route recommendation system using FastAPI and local LLMs to generate personalized pedestrian routes based on environmental data and user health profiles. Developed a dynamic scoring algorithm to promote safe, inclusive mobility within an interactive mobile application.

Contributed as a Data Engineer to a neuro-symbolic clinical decision support system for diabetes management. Developed robust data pipelines, including ETL processes for continuous glucose monitoring (CGM) sensor data and a streaming simulator for real-time risk alerts.

A personal portfolio built from scratch with Nuxt.js, featuring a dynamic canvas background, serverless functions on Vercel for Spotify API integration, and a fully responsive design.

A distributed in-memory data storage system built in Java, designed for high concurrency. The system features a multi-threaded, TCP client-server architecture with a custom binary protocol for data serialization. It implements robust concurrency control using thread pools, thread-safe queues, and `ReadWriteLock` for atomic operations, supporting simple, composite, and conditional (blocking) read/write requests.

A complete compiler for the standard Pascal language, developed in Python. This project encompasses all key compilation phases: lexical analysis for tokenization, syntactic analysis to build an Abstract Syntax Tree (AST), and a custom semantic analysis module to ensure type safety and program coherence. The compiler translates the AST into an intermediate representation and generates final machine code for a target Virtual Machine.

A web-based platform that leverages LLMs to help managers automate the process of getting a curriculum vitae of the team members, using only an excel with data, and a photo.

A secure client-server application for confidential document sharing. The system uses mutual authentication with X.509 certificates, ECDH for perfect forward secrecy, and a hybrid AES-GCM encryption model to ensure that not even the server can access file contents. Features a command-line interface for managing files, groups, and permissions.

A modular C++ 3D engine with a custom model generator and real-time OpenGL rendering via XML scene descriptions.