CV
For the full version, see my LinkedIn.
Experience
- Research Scientist, Meta Superintelligence Labs, Zurich (Oct 2025 – present)
- Media Generation team. Working on generative model distillation and real-time interactive world models, delivering 10× speedups for image-editing models deployed across Meta’s products.
- Visiting Researcher, Meta FAIR, Montreal (2024 – 2025)
- Developed reward-guided decoding to control multimodal LLMs at inference time, cutting object hallucination by ~70% at 6× the sample-efficiency of rejection sampling (ICCV 2025), advised by Michal Drozdzal and Adriana Romero.
- Research Scientist Intern, Meta FAIR, Montreal (2023)
- Built automatic prompt optimization (OPT2I) that improved text-to-image consistency by up to 25% while preserving image quality (TMLR 2024, Featured Certification), advised by Adriana Romero and Michal Drozdzal.
- Research Intern, Element AI, Montreal (2020 – 2021)
- Developed Seasonal Contrast (SeCo), a self-supervised pre-training method for remote sensing, improving land-cover classification by +8 mAP over ImageNet pre-training and matching full-label accuracy with 100× fewer labels (ICCV 2021; 500+ citations), advised by Pau Rodríguez and David Vázquez.
- Computer Vision Engineer, Mediapro R&D, Barcelona (2019 – 2020)
- Built real-time video analysis pipelines for automatic sports production, deployed in live broadcast.
- Research Assistant, Image Processing Group, UPC, Barcelona (2018 – 2019)
- Researched self-supervised visual representations for sample-efficient reinforcement learning, advised by Xavier Giró and Víctor Campos.
- Deep Learning Engineer, Restb.ai, Barcelona (2017 – 2019)
- Built and deployed production visual-recognition models for real-estate property imagery at scale.
- Research Assistant, Architectures and Compilers Group, UPC, Barcelona (2016 – 2017)
- Optimized deep neural networks for real-time inference on power-constrained edge devices (Nvidia Jetson TX1), achieving a 1.55× speedup, 31% lower energy, and an 86% smaller memory footprint, advised by Antonio González and José María Arnau.
Education
- Ph.D. Computer Science, Mila / Université de Montréal, 2021 – 2025
- Thesis: Towards efficient, reliable and measurable vision-language systems
- Supervised by Prof. Aishwarya Agrawal. GPA: 4.3/4.3.
- Nominated for the Dean’s Honour List.
- M.Sc. Computer Vision, Computer Vision Center / Universitat Autònoma de Barcelona, 2018 – 2020
- Rank 1/35. Grade: 9.58/10. Recognition of outstanding academic achievement.
- Thesis: Self-Supervised Visual Representation Learning for Remote Sensing.
- B.Sc. Informatics Engineering (Computer Science), Universitat Politècnica de Catalunya, 2013 – 2017
- Rank 1/235. Grade: 9.52/10. Recognition of outstanding academic achievement.
- Thesis: Adapting Deep Neural Networks to a Low-Power Environment.
Technical Skills
- Languages: Python (proficient); C++, Java, C, CUDA, SQL (familiar)
- ML Frameworks: PyTorch, HuggingFace (Transformers, Diffusers), vLLM, NumPy, OpenCV
- Tools: Git, Docker, LaTeX, TensorBoard, Weights & Biases, Submitit
- Systems: Linux/Unix, Slurm, distributed training/inference, parallel programming
Selected Awards
- Graduate Excellence Scholarships, Université de Montréal (J. Armand Bombardier, DIRO, J.A. DeSève, and Artificial Intelligence funds) (2022, 2023, 2024, 2025)
- National End-of-Degree Award in University Education, Spanish Ministry of Education (2021)
- Mila PhD Scholarship, Mila - Quebec AI Institute (2021)
- McGill Engineering Doctoral Award (declined), Faculty of Engineering, McGill University (2021)
- Mitacs Accelerate International Grant, Mitacs Canada (2020)
Academic Service
- Invited Talks: Deep Learning Barcelona Symposium 2025 — spotlight talk (recording)
- Reviewer: CVPR 2026, ECCV 2026, CVPR 2025, ICLR 2025, MAIS 2024, EMNLP 2024, ECCV 2024, CVPR 2024, NeurIPS 2023, ACL 2023, ICCV 2021
- Teaching Assistant: Links between Computer Vision and Language, UdeM (2023); Postgraduate AI for Deep Learning, UPC School (2020); Summer School on Deep Learning for Vision, UPC (2019)
Languages
- English (fluent, TOEFL 109/120), French (advanced, C1), Catalan (native), Spanish (native)