Senior AI Product Engineer · Full-Stack Engineer
I build production AI systems end to end—from retrieval and evaluation to APIs, workflows, and polished product interfaces.
Quito, Ecuador · Remote (LATAM) · Spanish (native) · English (fluent)
I have 11+ years of experience shipping production software across AI, legaltech, fintech, climate, healthcare commerce, ecommerce, and SaaS. My sweet spot is turning ambiguous product problems into reliable systems—then measuring, hardening, and operating them in production.
I work across the entire product stack: LLM and retrieval pipelines, Python services, workflow orchestration, data models, React interfaces, cloud infrastructure, and CI/CD.
I'm building production AI systems for complex legal-document workflows:
- Ingestion, scoped hybrid retrieval, and grounded generation orchestrated with Temporal DAGs, retries, and human approval gates.
- MCP-powered document search and chat using pgvector, BM25, reranking, reciprocal-rank fusion, and source-level traceability.
- Automated LLM evaluation with golden sets, structural and citation checks, LLM-as-judge, plus DOCX/OOXML generation that preserves formatting and provenance.
- Architected a multi-tenant FastAPI and PostgreSQL platform that reduced infrastructure costs by 40% ($150K–$250K annually).
- Led a team building a recommendation system that combined business rules, statistical ML, foundation models, and React/Next.js experimentation.
- Scaled a commerce platform supporting 500+ medspas and $1M–$2M in weekly GMV.
- Led engineering teams of up to six engineers across backend and frontend delivery.
| Project | What it demonstrates |
|---|---|
| Clara | Full-stack marketplace with Gemini, Stripe, background jobs, PostgreSQL, and React/Next.js product interfaces. |
| ClearSky AI | Hackathon-winning flight-safety assistant combining 12+ deterministic risk rules with Gemini and live web intelligence; built with FastAPI and Next.js. |
| Kerly | Nutrition-first AI meal assistant backed by Python, PostgreSQL, Redis, and Celery. |
| RoleLoop | Evidence-first job-search workflow with layered Python architecture, persistent workflow state, immutable Git-backed evidence, a terminal UI, and a headless CLI. |
| Nova | Reusable full-stack foundation with Python services, Next.js, GraphQL/REST, PostgreSQL, Redis, Celery, CI/CD, and optional SwiftUI. |
AI systems: RAG, hybrid retrieval, semantic search, reranking, agents and tool use, structured outputs, MCP, and automated evaluation/monitoring.
Product engineering: Python, FastAPI/Django, Temporal, PostgreSQL/pgvector, TypeScript, React/Next.js, AWS, Docker, and GitHub Actions.
- M.S. in Computer Science, The University of Texas at Austin.
- Python community speaker at PyCon Panama, Python Guatemala, and PyTexas.
- 71k+ Stack Overflow reputation and a Python Gold Badge.
- Winner of an AI aviation hackathon and an OSS hackathon for a Cal.com message-queue integration.
If you're building an AI product that has to be useful, traceable, and dependable in production, let's talk.