I'm an engineer who likes building solutions to problems. A builder, basically.
Flight-data processing with prediction accuracy and ICAO KPI14 punctuality analysis.
A full-stack system that ingests flight data, compares predicted with actual flights, measures punctuality against ICAO KPI14 tolerance windows, and scores trajectory accuracy.
Overview.
Trajectory accuracy (sample data).
Flight search (sample data).
Full analysis page (sample data).
Every screen uses sample data.
A drag-and-drop editor for multi-step AI agent workflows, with live per-node runs.
Live video mock interviews with an AI interviewer, then a written review of every answer.
A retrieval-augmented chatbot over Stanford's Black community archives.
Researchers ask plain-language questions; the system retrieves the relevant passages from long scanned and digital records and answers from them.
An AI research pipeline that reads companies' sustainability reports and scores them.
An order, payment and shipping lifecycle built on Temporal workflows.
Durable order orchestration: workflows, activities, signals, timers, retries and child workflows, with idempotent payments, separate task queues and an API to start workflows, send signals and inspect live state.
AI journaling: log your mood, get feedback on each entry, and talk it through.
Talk about your day in a voice note; it reads the emotion in your voice and tracks your mood.
A voice journal: record a short note about your day, and the app picks up the emotion in your voice, tracks your mood over time and gives you personal insights.
I build multi-agent LLM pipelines that review long, high-stakes documents, with evals and checks built in, as a forward deployed engineer at a Y Combinator-backed startup.
Before that I was a solutions engineer at Chevron, interned at Embraer and at Stanford, and studied computer science at UC Davis.
On the side, I build software and AI tools for companies. I work in Spanish and English.
For a role, a project or a hello, in English or Spanish.
