Spain
I build end-to-end software products and data systems — from architecture and APIs to production infrastructure — with an AI-native working style: design contracts and acceptance criteria first, accelerate implementation with modern AI tools, and keep full ownership of review, security, debugging, and deploy decisions. With AlfaTactix I took a complex SaaS from idea to production alone: React/TypeScript and FastAPI, PostgreSQL/Redis, Docker on Oracle Cloud + Vercel, Stripe and Spanish e-invoicing, auth/security (JWT, 2FA), and bilingual SEO at scale. Along the way I designed a Python backtesting stack (NumPy/Pandas, multi-timeframe execution, risk and performance analytics) and a documentation-grounded RAG architecture (pre-release). Core stack I work with Frontend: React, TypeScript, Vite, MUI, Tailwind, Formik/Yup/AJV, i18next, DnD Kit Backend: Python, FastAPI, SQLAlchemy, Alembic, Pydantic, Redis Streams, JWT/RBAC Data & ML: PostgreSQL, NumPy/Pandas/SciPy, scikit-learn, time-series & geospatial pipelines AI: RAG design (chunking, embeddings, pgvector, retrieval-first grounding), LLM-agnostic generation layers Cloud: Docker Compose, OCI, Vercel, Caddy, monitoring & structured logging Research background As a Postgraduate Researcher at the University of Seville (2021–2024), I built automated remote-sensing pipelines for Copernicus (Sentinel) and Landsat data, geospatial ETL, and predictive models (Random Forest, regression) for drought and environmental risk signals with strict time-based validation. I hold an M.Sc. in Civil Engineering (Hydraulic Structures & Water Resources) and have peer-reviewed publications in coastal aquifer / hydro-environmental modelling. How I approach problems I like turning messy domains into structured products: inventory sources, separate what is measurable from what is oversold, and ship something decision-ready. Example: for an xNova take-home on Middle East trade data I evaluated 61 sources, produced a full research package and pilot recommendation, and built an interactive prototype (middle-east-opportunity-explorer.vercel.app). What I’m interested in 0→1 and 1→n product engineering, AI systems that stay grounded in real data (RAG, validation, honesty about limits), fintech/SaaS platforms, and applied geospatial or environmental intelligence where research rigor meets scalable software.
Live product: https://alfatactix.com | API: https://api.alfatactix.com Portfolio proof of end-to-end ownership — seeking full-time roles (product in maintenance / non-competing mode). • Designed, architected, and shipped AlfaTactix from zero to production as a solo builder (AI-native delivery: architecture-first, AI-accelerated implementation, human-owned review and production decisions). • Product: 6-step visual no-code Strategy Builder · User dashboard · MQL5 code generator · Admin panel · Subscription & checkout · Role-based access control • Code generation: JSON strategy architecture · MQL5 pipeline (Parse → Transform → Build → Output) with dynamic input parameters (rule/pipeline-based transpiler — not LLM code generation) • Backtesting: Python engine (~75 modules: multi-timeframe, parallel execution, indicators/filters, risk, portfolio, analytics) + FastAPI job API — feature-complete in codebase; public product release pending • RAG (designed / pre-release): documentation-grounded Knowledge Assistant architecture (chunking, multilingual embeddings, PostgreSQL + pgvector, retrieval-first prompting, source attribution) • Content & SEO: ~300+ indexed URLs · bilingual Academy (EN/ES) · automated sitemap, hreflang, Playwright prerender • Payments & platform: Stripe subscriptions · Redis Stream webhook worker · invoice PDF · Spanish Facturae XML (XAdES) · VAT/compliance path • Security & infrastructure: JWT + refresh · TOTP 2FA · CSRF/HSTS/rate limiting · Docker Compose · Oracle Cloud (OCI) · PostgreSQL · Redis · Prometheus/Grafana · Vercel • Traction (first ~3 months): 100+ organic registrations · 541K Google Search Console impressions
Postgraduate research in environmental data science at the Department of Geography — turning large satellite and climate datasets into validated predictive signals for regional monitoring in Andalusia. • Built automated Python pipelines for Copernicus (Sentinel) and Landsat imagery: extraction, cloud masking, reprojection, and spatial harmonization • Engineered hydro-climatic and vegetation-stress indices (SPI, SPEI, SSMI, SSFI and related signals) from high-density spatial–temporal data • Assembled spatial–temporal panel datasets and trained/compared models (Random Forest, linear/ridge regression) for environmental risk prediction • Applied strict time-based hold-out validation to control leakage from spatial and temporal autocorrelation • Collaborated with geographers and ecologists to translate model outputs into decision-support insights Stack: Python, Pandas, NumPy, SciPy, scikit-learn, GeoPandas, Rasterio, GIS, Jupyter