About Me
Pharmacist by training who codes Machine Learning for drug discovery and pharmacogenomics. My profile combines clinical drug knowledge with the technical ability to design genomic analysis pipelines, model gene–drug interactions with graph neural networks (GNNs), and build the software that supports them.
I aim to contribute to pharmaceutical R&D, clinical bioinformatics or computational drug development teams, bringing a cross-disciplinary perspective between biology, chemistry and data science. Self-taught, always learning, and open to research and PhD/CDT collaborations.
Technical Skills
Programming & Data Python · PyTorch · Pandas · Polars · NumPy · Scikit-learn · C++ · Bash · SQL · Git
Machine Learning & Deep Learning Neural Networks · GNN (GATv2) · PyTorch Geometric · DeepFM · Transformers · Multi-Task Learning · Optuna · Cross-Validation · Focal Loss
Bioinformatics & Genomics BWA-MEM · GATK · Samtools · Picard · Freebayes · VCFtools · Ensembl VEP · FastQC · FastP · NGS Pipelines
Computational Pharmacology Pharmacogenomics · PharmGKB · ClinPGx · dbSNP · RDKit · Molecular Graphs · Molecular Docking · ADMET · Structure-Activity Relationship
Software Engineering & MLOps Pydantic v2 · FastAPI · pytest · ruff · CI (GitHub Actions) · Typed design · REST APIs
Infrastructure GNU/Linux (Debian) · WSL · Hyprland/Sway · Quarto · uv · Git
Education
- Master’s in Bioinformatics — Valencia International University (2025–2026)
- Degree in Pharmacy — University of Granada — Focus on Pharmacogenomics
- Python for Data Science — Universitat Oberta de Catalunya (300h)
- ML & Big Data for Bioinformatics — University of Granada (100h)
- Nanomedicines — Pharmacy Academy Santa María de España
Languages
- Spanish — Native
- Tamazight — Native
- English — B2 (Professional proficiency)
Featured Projects
Pharmagen — AI for Pharmacogenomic Prediction
End-to-end platform integrating a complete bioinformatics pipeline (FASTQ → VCF → Annotation) with a Deep Learning model to predict a patient’s phenotypic response to specific drugs.
The project is evolving on two fronts: v1 (a research prototype, DeepFM + Transformer) that validated the hypothesis, and v2 (in progress), migrating to a Two-Tower GATv2 graph architecture and hardening the codebase to industry standards (Pydantic v2, FastAPI, CI) with a rigorous, leakage-aware evaluation protocol.
Stack: Python 3.14 · PyTorch · PyTorch Geometric (GATv2) · RDKit · Optuna · FastAPI · Pydantic v2 · BWA · Freebayes · Ensembl VEP
Durin — Clinical Software for Polypharmacy (early development)
Tooling to support the care of polymedicated geriatric patients, where the risk of drug–drug interactions and adverse effects grows combinatorially with each added medication. It ingests a medication list and clinical profile and aims to surface the risks worth a clinician’s attention.
Foundation already built: typed domain models (Pydantic v2), a custom ATC/DDD catalog via scraper, and internationalization (ES/EN). On the roadmap: the interaction/risk engine and an AI layer to prioritize and explain risks. Not a medical device; no clinical use.
Stack: Python 3.14 · Pydantic v2 · FastAPI · BeautifulSoup · Polars
Other Repositories
- Portfolio (this site) — Bilingual website built with Quarto on Debian; see the repository.
- dotfiles_Aderfi — Hyprland/Sway environment configuration for Debian and WSL. Reflects Linux administration proficiency.
Blog
Posts on engineering decisions, deep learning architecture analysis and reflections on applied bioinformatics.