About
Data Science & AI student at Universidad Politécnica de Madrid (UPM), focused on shipping production-grade ML and LLM systems end-to-end. Currently architecting an AI research intelligence platform at UPM as an AI Software Engineer & Architect, and contributing to the open source projects that power the ML ecosystem.
Passionate about LLM fine-tuning, multi-agent orchestration, and making ML testing rigorous.
Open Source Contributions
Contributor to PyTorch (Meta), Hugging Face Transformers, OpenAI Triton, Apache TVM, NVIDIA TensorRT-Model-Optimizer, Lightning AI, vLLM, OpenAI Agents Python, Tauric Research TradingAgents, and pysystemtrade.
Featured Projects
- TrueRisk: Climate emergency management platform with ML-powered risk scoring for 52 Spanish provinces. 2nd place at Cubepath 2026 Hackathon. (FastAPI, XGBoost, LightGBM, LSTM, TFT, Next.js)
- FractalSig: Hybrid JAX/PyTorch generative model for rough paths with a learned Besov-wavelet decoder and differentiable IDWT. (JAX, PyTorch)
- YuhoLens: Span-cited English investor memos produced by a full-parameter 14B Qwen fine-tune on a single AMD Instinct MI300X; 4-agent LangGraph pipeline. (Python, LangGraph, vLLM)
- checkllm: A pytest plugin and CLI for testing LLM-powered applications with deterministic checks and LLM-as-judge evaluation. (Python)
- Quant Finance Engine: Event-driven backtesting system with iTransformer forecasting, InvAD crisis detection, and NeuralHRP allocation. (Python, PyTorch, XGBoost)
Education
Bachelor's Degree in Data Science and Artificial Intelligence — Universidad Politécnica de Madrid (UPM). Honors (Matrícula de Honor) in Data Science Programming and Natural Language Processing.
Certifications
- AWS Certified AI Practitioner — Amazon Web Services
- Azure AI Fundamentals — Microsoft
- Financial Markets — Yale University (Risk Management & Behavioral Finance)
Technologies
Python, PyTorch, JAX, XGBoost, Scikit-learn, Pandas, NumPy, LangChain, Docker, AWS, Azure, Git, React, Next.js, Tailwind CSS, Three.js, SQL, Bash