Applied AI and machine learning researcher at Instituto de Telecomunicações, and PhD candidate with a background in Electronics and Telecommunications Engineering.
My work combines deep learning and signal processing with LLM-based systems for scientific document analysis and engineering workflows. I develop and evaluate these systems with attention to experimental integrity, traceable evidence and explicit limitations.
| Project | Focus |
|---|---|
| Drone RF Classifier | A compact PyTorch CNN for raw-IQ signal classification, with recording-disjoint splits, a synthetic demo and documented experiment provenance. |
| Antenna Paper Extraction | Agent-based extraction of antenna architecture from scientific papers, producing evidence reports and execution traces. The experimental MCP workflow is available on the development branch. |
| MCP PDF Ingestion | A PDF evidence server adapted from an existing reviewer project, exposing deterministic retrieval tools and explicitly requested visual inspection through MCP. |
| Project | Current focus |
|---|---|
| CST MCP | Extending and validating an existing MCP server for CST Studio Suite. Current work focuses on parametric geometry, capability testing and reliable automation of engineering workflows. |
| RepoLens | Very early-stage project. Initial development focuses on Python foundations, data modelling and Git history retrieval. Repository change analysis is the longer-term goal; broader analysis and AI capabilities are not yet implemented. |
- Machine learning: PyTorch, TensorFlow, model evaluation and reproducible experiments.
- Signal processing: RF sensing, raw-IQ data and signal classification.
- AI systems: LLMs, multimodal models, agent workflows and Model Context Protocol.
- Software development: Python, testing, command-line tools and Git.