Education, experience, and skills.
Education
Universitat Autonoma de Barcelona (UAB)
Sept 2026 - present
Covers the full AI spectrum: strong math and theory, cognitive science, and applied engineering. Project-based learning.
La Salle Campus Barcelona (Universitat Ramon Llull)
Sept 2025 - June 2026
Systems-level focus: C programming, computational infrastructure, algorithms, data structures, probability and statistics, calculus, linear algebra, and computational logic. Top of class grades.
DeepLearning.AI / Stanford (Coursera)
July - Aug 2025
Andrew Ng's three-course specialization: supervised learning, advanced algorithms, and unsupervised learning.
IU International University of Applied Sciences
Sept 2024 - June 2025
Applied AI curriculum: machine learning fundamentals, Python, mathematical analysis, and hands-on project work. Completed alongside full-time employment.
Sabadell, Barcelona
Sept 2022 - June 2024
Spanish university entrance qualification. Completed while working full-time.
Work Experience
Hydac Technology, Barcelona
Sept 2022 - Sept 2025
Started in warehouse operations handling inventory, shipping, and receiving. Promoted to the logistics office after 6 months, managing order processing, stock coordination, and supplier communication. Worked full-time alongside Bachillerato and first year of university studies.
Technical Skills
Selected Projects
Fine-tuned DistilBERT with LoRA for binary sentiment analysis. 0.3% of parameters, ~1 MB adapter. Live demo on HuggingFace.
Full reimplementation from He et al. 2016 paper: BasicBlock, residual connections, batch norm, and He init all written by hand. Trained on CIFAR-10.
Unsupervised anomaly detection on 284K transactions (0.17% fraud). Compared Isolation Forest, LOF, and One-Class SVM.
Tabular reinforcement learning agent that learns to balance a pole through 10,000 episodes of trial and error.
Time-series prediction of Barcelona bike-sharing availability with temporal features, lag engineering, and LightGBM.
Fine-tuned SmolLM2-360M as a machine learning tutor with 40 curated Q&A pairs and chat templates.
Predicts student grades from study habits, sleep, and attendance. Compares 6 regression models.
K-Nearest Neighbours built from zero with NumPy. Euclidean distance and majority voting, no libraries.
CLI habit tracker with streak tracking, struggle detection, and full pytest suite. Built with Python and SQLite.
Languages