I build and operate reliable data platforms
Data engineer across cloud, DevOps and AI.
Python and Databricks at the core, production systems end to end.
Built at Amadeus BASF BBVA AI Factory Adaltas Banco Nacional de Costa Rica
2026–now
2025–2026
2023–2025
2022–2023
2020–2022
A tiny AI coding agent built from scratch, to show what Claude Code, Cursor and Codex actually do: ask the model, run the tools it asks for, send back the results. Each concept is one commit, with its own docs site walking through every stage.
Lets each workspace team audit its Hive metastore and migrate tables and grants to Unity Catalog on its own, instead of a central team migrating workspace by workspace. Built for a large multi-workspace Databricks estate.
Declarative ingestion from SAP, Kafka, and REST APIs into Delta Lake, governed through Unity Catalog. Config-driven, retry-aware, schema-validated. I led the team that built it.
End-to-end orchestration for analytics workloads on Databricks with Unity Catalog governance and automated quality checks.
Retrieval-augmented generation for a Financial Plan creation chatbot, from document retrieval to grounded generation, for a banking data team.
Also from scratch: kafka-python, a Kafka broker implementation written to understand the protocol rather than just use a client library.
Engagements start with a short conversation, then run as discovery, technical design, build, and handover with documentation your team can own.
Architecture and implementation of cloud-native data platforms on AWS and Azure. From ingestion to serving, built for scale and maintainability.
Move Hive metastore estates to Unity Catalog: audit what exists, plan catalogs and grants, and give workspace teams tooling to migrate on their own.
Robust ETL/ELT pipelines with PySpark, Databricks, and cloud-native tooling. Schema management, retries, idempotency baked in.
Automated training, testing, and deployment pipelines. Containers, IaC, and workflow orchestration for repeatable delivery.
Metrics, logging, alerting, and SLOs for data systems. Keep pipelines healthy and catch problems before users do.
Optimize jobs, clusters, and storage layers for throughput and cost. Practical improvements, not theoretical benchmarks.
I've built data and ML systems in consulting, banking, chemicals, and travel tech across Costa Rica, France, and Spain: Hadoop clusters on AWS, RAG pipelines, and Databricks platforms governed by Unity Catalog. A background in mathematics (degree and master's, Costa Rica) shapes how I work: explicit contracts, idempotent pipelines, and observability that tells you something broke before your users do.
Today I work independently with teams that need a senior engineer who can design the platform and also run it.
Tell me about the platform, the team, and what is getting in the way. I usually reply within 24 hours.
