Quantitative developer & energy economist · PhD candidate at UPV/EHU · Bilbao, Spain
I study how volatility moves between European power markets and publish the tools behind that research as open-source Python. I also build production data and AI products through Zero Sum.
📫 Available for freelance projects, hourly or milestone-based. See Work with me.
Electricity Market Integration under Stress: Volatility Spillovers and the Iberian Exception. Accepted for presentation at CFE-CMStatistics 2026.
Realized (co)variances from hourly ENTSO-E day-ahead prices, MHAR-LASSO dynamics, and Diebold–Yılmaz connectedness via generalized FEVD.
| Project | What it does |
|---|---|
| connectedness · PyPI | Diebold–Yılmaz connectedness, realized (semi)covariances, MHAR-LASSO, static & rolling spillover indices |
| loadcast · Live dashboard | Day-ahead electricity load forecasting for Spain, Germany & France: XGBoost vs LSTM vs a from-scratch Transformer, benchmarked against the TSO, with live daily forecasts |
| entsoe-data-toolkit | Download, cache, process and visualize ENTSO-E electricity market data |
| VoxLab | Local-first AI voice lab: multi-character dialogues, voice profiles, audio style presets |
- Forecast: a forecasting SaaS covering ARIMA, gradient boosting, deep learning (N-HiTS, PatchTST) and foundation models (Chronos, TimesFM) on serverless GPUs.
- Invoice AI: an LLM-based invoice reader that books supplier invoices into an ERP, with human approval.
Quant & ML: time-series econometrics, realized volatility, HAR/VAR, LASSO, forecasting, deep learning, LLM applications
Engineering: Python, SQL, TypeScript, FastAPI, Django, Next.js, PostgreSQL, Docker, Solidity/Foundry
Domains: EU & Iberian electricity markets, natural gas, commodities
I'm open to freelance work, on an hourly basis or as fixed-price milestones: forecasting and time-series models, energy-market data pipelines, quantitative research, and LLM-based data extraction.