Representation Learning
Visual embedding spaces that capture semantic structure, variability, and transferable information.
Profile
I am a dual-degree PhD student at Carnegie Mellon University and Instituto Superior Técnico. My PhD is supervised by Fernando De la Torre at Carnegie Mellon University and by Carlos Santiago and Catarina Barata at Instituto Superior Técnico. My research interests lie in computer vision, particularly generative models, representation learning, and self-supervised learning. I am interested in how visual models can learn structured and transferable representations from unlabeled data, and how those representations can support more controllable and semantically meaningful generation.
My background spans probabilistic state estimation, prototype-based learning, and applied machine learning. Before starting my PhD, I was a Visiting Scholar at Carnegie Mellon University’s Physical Perception Lab. My recent work includes vMFProto, a distributional prototype-learning approach for interpretable image classification, as well as production AI systems developed at Pairwire and GLS.
Broadly, I want to understand how visual models organize information in their latent spaces, how useful representations can be learned with limited supervision, and how those representations can be used for generation.
Visual embedding spaces that capture semantic structure, variability, and transferable information.
Meaningful and robust visual representations learned from unlabeled data.
Structured, controllable, and semantically meaningful visual generation.
Beginning doctoral research in computer vision through the CMU Portugal Program.
Distributional part prototypes for interpretable image classification.
Worked with Prof. Shubham Tulsiani on visual representation learning.
Research output
A distributional part-prototype framework using von Mises-Fisher mixtures and structured optimal-transport assignments to improve explanation quality.
Academic and professional path
2025 — 2026
Institute for Systems and Robotics · Lisbon
Distributional prototype learning and structured representation learning in self-supervised visual feature spaces.
2025
Physical Perception Lab · Carnegie Mellon University
Worked with Prof. Shubham Tulsiani on visual representation learning.
2024 — 2025
Pairwire · Lisbon
Led NLP, retrieval-augmented generation, and graph reasoning systems for consulting workflows.
2023 — 2025
General Logistics Systems · Lisbon
Built customer classification and parcel anomaly systems for GLS Portugal and Germany.
2022 — 2023
Institute for Systems and Robotics · Lisbon
Developed non-Gaussian sensor-fusion and state-estimation methods for distributed surveillance.
2026 — present
Carnegie Mellon University · Instituto Superior Técnico
Dual-degree doctoral program through the CMU Portugal Program, with research interests in computer vision, representation learning, self-supervised learning, and generative models.
2022 — 2024
Instituto Superior Técnico · Lisbon
Control, robotics, and AI. Thesis on observer design for non-Gaussian noise.
2018 — 2021
Instituto Superior Técnico · Lisbon
Mechanics, dynamics, and control systems.