01 / 08
OpsPilot AI
2026A human-in-the-loop agent that investigates incidents and drafts postmortems, gated by a reproducible eval suite of 30+ cases.
- LangGraph
- FastAPI
- Langfuse
- Grafana
Repo
Runs on a local observability stack
Business constraints into systems people can rely on — AI included.I turn business constraints into systems people can rely on — AI included.
Fullstack Developer & AI Engineer in Mexico City. Scope with Product, architecture decisions, frontend and Go services — plus the reproducible evaluation, tracing and guardrails that anything with AI in it has to pass.
Selected work
03
LegalMove Pro
2025Compares a contract amendment against its original and returns structured changes, risk flags and human-review recommendations.
- Go
- Next.js
- Azure
- Terraform
Repo
No live demo — see the repo and the walkthrough above
01 / 06
FitTrack AI
2025A cloud-native app with persisted AI recommendations and direct-to-storage uploads, backed by 100 backend and 319 client tests in CI.
- Flutter
- FastAPI
- Azure OpenAI
- Terraform
Repo
No live demo — mobile app, see the repo and the walkthrough above
How I work
The part that decides whether an AI feature survives contact with real users.
Reproducible evaluation
Every change is measured against the same gate before it ships — ambiguous signals, insufficient evidence and adversarial input included.
Tracing and observability
Each decision is traceable end to end, through LLM traces and an operational stack of metrics and logs.
Guardrails and structured output
Models return shapes the system can validate, not prose the system has to guess at.
Human in the loop
The output is routed to a person to approve, not passed off as a final answer. Approvals stay with the human.
Cost and latency control
Inference is a line item and a response time, and both are budgeted before the feature ships.
About
I have been a Fullstack Developer at Kunzapp in Mexico City since July 2023. I work with Product to shape scope and tickets, make architecture decisions, build frontend features, Go services and AWS workflows, and review the team's PRs. Services I built with Go and PostgreSQL reduced average API response time by 25%.
I also built a Vue Chrome extension integrated with Google APIs that grew beyond 1,000 active users. Before moving fully into engineering, I spent five years across data and project management: I automated SQL reporting at Bestseller LATAM and cut reporting time by 30%, after two management and analysis roles at El Palacio de Hierro.
My education connects both sides: Full Stack at Henry (2022–2023), a Master's in Project Management from UNITEC (2020), a Bachelor's in Business Administration from UNITEC (2014), and Henry's AI Engineering Certificate (2026). I am now looking to build dependable products where AI earns its place through reproducible evaluation, observability, and business judgment.