I'm a Machine Learning Engineer at Avature and a PhD in AI / Data Science & Engineering from the University of Seville.
My work sits at the intersection of applied machine learning, time series, generative models and production-oriented AI systems. I have worked on forecasting, imputation, medical AI, renewable energy, industrial anomaly detection and autonomous agent evaluation.
Right now, I am especially interested in:
- Reliable LLM agents, evaluation frameworks and personal AI systems.
- Time-series forecasting, missing-data imputation and spatio-temporal modelling.
- Generative modelling, diffusion/consistency models and practical ML research.
- Applied AI for energy, health and biotech products.
- ML Engineering: building and deploying machine-learning systems in real products.
- Agent evaluation: creating reproducible ways to benchmark LLMs and autonomous agents.
- Research-to-product: turning research ideas into useful tools, not just notebooks.
- VenpraLab: applying biotechnology and data-driven thinking to biomethane and anaerobic digestion.
| Project | What it is | Topics |
|---|---|---|
| personal_agent_eval | Open evaluation framework for LLMs and autonomous agents, with deterministic checks and judge-based scoring. | LLMs, agents, evals, OpenClaw |
| CoSTI | Consistency Models for multivariate time-series imputation. | Time series, imputation, generative models |
| GAIN-Pytorch-Lightning | PyTorch Lightning implementation of Generative Adversarial Imputation Networks. | Missing data, GANs, PyTorch |
| sepsis-review | Deep-learning framework for sepsis prediction with MIMIC-III data. | Medical AI, tabular data, TensorFlow |
| electric-multivariate | Electricity price forecasting experiments with univariate and multivariate deep-learning models. | Forecasting, energy, deep learning |
- Time-series forecasting and imputation.
- Diffusion, consistency and generative models.
- NLP and LLM-agent evaluation.
- Applied ML for renewable energy, health and industrial systems.
You can find my publications on Google Scholar. Recent work includes:
- Consistency models for spatio-temporal imputation.
- Day-ahead electricity price forecasting with deep learning.
- Spanish electricity price forecasting with lagged and non-lagged models.
- Sepsis prediction frameworks using MIMIC-III.
- Semi-real-time solar irradiance forecasting.
Languages: Python, SQL, JavaScript/TypeScript, Kotlin
ML: PyTorch, TensorFlow/Keras, PyTorch Lightning, scikit-learn, pandas, NumPy
Systems: Docker, GitHub Actions, FastAPI, PostgreSQL, Linux
Focus areas: LLM agents, evaluation, time series, generative models, applied ML