Work
Public
DOMA
Urban planning review product for architecture firms in Chile: case-file upload, two-phase LLM analysis, live progress narrative, and professional PDF reports.
Public
★ Chief Viber · Plotly Analytics Vibe-a-ThonOECD Pensions Explorer
An interactive explorer of OECD pension-system trade-offs: it connects replacement rates, spending, retirement age, and old-age poverty over a reproducible dataset. Won the Chief Viber prize at the Plotly Analytics Vibe-a-Thon and was published in Plotly’s official examples.
Public
Chile Macro Forecasting
Reproducible system for forecasting Chilean macro series from the Central Bank BDE API: model pool, rolling backtests, DuckDB history, and a Dash/Plotly dashboard.
Public
DIPRES Evaluation Bridge
Auditable system connecting evaluated Chilean public programs with budget lines, execution data, and documentary evidence, treating each bridge as a reviewable hypothesis.
Public
Chile Hospital Map
Data product for mapping Chilean public hospitals, health services, location, and annual bed capacity from 2020-2024 with reproducible ETL, API, and a D3 map.

Public
★ 1st place · Plotly ChallengeMTA Ridership
A focused app that explains NYC transit recovery with three key views, sober UX, and a clear narrative. First place in the Plotly Holiday Season App Challenge.
Private (redacted)
Public procurement tooling (redacted)
Private engagement: pipeline + app that turns tenders and awards into decision-ready data. Screens are redacted; no names or amounts.

Public
Plotly in production (Pokémon Dashboard)
An interactive dashboard built with Python + Plotly to explore and compare data, with practical lessons on performance, UX, and shipping without writing JavaScript.

Public
TARCJA: judicial document extraction & dashboard
End-to-end application of a general-purpose LLM taxonomist to a corpus of judicial appeal documents. Extracted structured records at scale, then built a static Plotly dashboard for exploration and multi-run comparison.
Public
Multiagent document taxonomist
General-purpose cloud-native platform for extracting structured data from unstructured PDFs. Induces an extraction schema from example documents, then processes batches through event-driven LLM workers at scale.

Public
Hackathon landscape analysis
Python pipeline that scraped, structured, clustered, and scored 4496 projects from the Gemini API Developer Competition on Devpost. Findings published as a LinkedIn post.
Available
Your project here?
If you need to turn scattered data into a useful tool, start with a short brief.