Backend engineer building event-driven Python systems, and Data Science student at UNSAM (Buenos Aires).
I design and run systems that have to survive contact with production: real traffic, real failures, real money. Right now that means a live trading platform I built and operate, executing client strategies on Hyperliquid with real capital.
Production trading platform (private, client project)
- Async Kafka workers + FastAPI services, PostgreSQL (SQLAlchemy/Alembic), Redis
- Saga-based distributed transactions coordinating multi-wallet fund management, with safe rollback/compensation
- Idempotent, self-healing consumers. No duplicate or lost orders across failures/restarts
- 1,500+ pytest tests covering order flows, failure paths, saga compensation
- Migrated the execution layer from Solana to Hyperliquid, rebuilt around its perpetuals DEX and WebSocket fill monitoring
- buscasam: Hybrid semantic + full-text search platform for the UNSAM community. Built with a teammate.
- polyglot-dwh: PostgreSQL/Supabase OLTP (3NF) feeding a Kimball star-schema warehouse via incremental Python ETL on GitHub Actions. pgvector semantic search, Redis (autocomplete, JWT blacklist, rate limiting), Power BI dashboards.
- potrerillos-wind-forecast: 12h-horizon wind forecasting. Compared SARIMA, LassoCV, LSTM under walk-forward evaluation, 28% RMSE improvement over a seasonal-persistence benchmark.
- quant-strategies: Config-driven backtesting framework on vectorbt with Optuna Bayesian hyperparameter search over QuestDB time-series data.
Backend & distributed systems: Python · FastAPI · asyncio · Kafka · PostgreSQL · Redis · SQLAlchemy/Alembic · Docker Data & ML: pandas · scikit-learn · statsmodels · TensorFlow/Keras · vectorbt · Optuna