I build AI systems that can be audited: LLMs for banking, calibrated models for finance, and data for cities.
Final year at UPV, now at IMT Atlantique. Based in València, always open to new projects and to working in any environment.
Every line starts at UPV. Select a station to read it.
AI Engineer, generative AI for wealth management
Jun 2026 – now
Sole technical owner and client contact for a retrieval system serving tier-1 banks. Also prototyped an LLM agent for portfolio-risk management, where every output must be traceable.
Research, client work and side projects. Real data, measured results, and models that report how sure they are.
We treated market prices as particles moving through a medium and asked when their diffusion regime changes. A multi-task Conv1D + BiLSTM network, trained on 100k synthetic trajectories, locates the changepoint and classifies the regime at the same time.
On S&P 500, IBEX 35 and crypto data it matched 12 of 15 known stress events, up to 45 days early. A Bayesian uncertainty layer turns each detection into a credible interval and a cost-sensitive action, so a person reviews the output instead of trusting a bare score.
Code private. Paper in preparation.
I own this system end to end as the only technical lead and the client's point of contact. In banking every answer has to be traceable to a source document, so evaluation is built into the pipeline from the start.
Neighbours and community gardens report problems on Telegram, by text or voice. Whisper transcribes, Gemma (via Ollama) extracts the who, where and how urgent, and the result lands in Supabase and a Neo4j graph as tasks with owners.
We went looking for real, messy data. Field sensors plus AEMET and SIAR weather records gave us five-day soil-moisture forecasts, showing a farm could run fewer sensors.
A habit design and tracking app with real-time visualisation. React and Supabase.
2025AutoReservas, a Telegram booking bot, and a UPV gym booking script with requests and Selenium. Notifications and credentials kept out of the code.
2024 – 25Work, community roles and study, most recent first.
| Jun 2026 – now | AI Engineer GPTadvisor, generative AI for wealth management. RAG delivery for tier-1 banks and a prototype LLM agent for portfolio-risk management. | AI & Finance |
|---|---|---|
| Mar 2026 – now | Data Scientist intern Generalitat Valenciana. Geospatial base for a multimodal transport network of the València metropolitan area. | Urban |
| Jan – Jul 2026 | Lovable Campus Leader UPV. Workshops, build nights and hackathons; co-organised the Valencia BuildJam. | Community |
| Sep – Nov 2025 | Data Scientist intern Generalitat Valenciana. Congestion-prediction prototype for Paiporta from 100k+ traffic and sensor records. | Urban |
| Sep 2024 – now | Student mentor ETSINF, UPV. Helping first-year students settle into university. | Community |
| Sep 2023 – now | Board member & project manager Sigma Data Club UPV. Workshops for 140+ members and a GenAI Hackathon with Lovable, GPTadvisor and MyInvestor. | Community |
| Until 2025 | Club projects UPV Investment Club. | AI & Finance |
| Sep 2026 – Jun 2027 | MSc Data Science IMT Atlantique, France. Erasmus+. |
|---|---|
| 2023 – 2027 | BSc Data Science Universitat Politècnica de València. Top Academic Performance Award, 2025. |
| Jul 2025 | University extension diploma in AI Samsung Innovation Campus. |
Five years ago I didn't know if I'd study medicine, computer science or biology. I followed my instinct into data science, and the path cleared as I walked it.
In 2025 UPV gave me its Top Academic Performance Award. What I enjoy most is work with real people on the other end: a farm-tech founder sharing his sensor data, a city planning its transport, a bank that needs every answer traced to a source.
Outside work I train, read, and keep one stoic rule close: do what depends on you, and accept what doesn't.
Always open to new projects, collaborations and roles, remote or on site. Email is the fastest way to reach me.