GovTech · Municipal Management · Technical Lead · Backend · Architect

CAPCA — Predicting the limits of a municipality.

A predictive platform that tells a Chilean municipality how many months remain before its basic services hit capacity — built with Universidad Católica de Chile, live in 11+ communes today.

Role Technical Lead · Principal Backend · Architect
Year 2020–2021
Team 11 people, multidisciplinary
Partner Universidad Católica de Chile
Funding FONDEF-IT · ANID

CAPCA — official presentation, Universidad Católica de Chile.

At a glance

10

Core municipal services monitored per commune

11+

Chilean municipalities in production today

5 yrs

Projection horizon per indicator

What CAPCA is.

CAPCA — short for Capacidad de Carga ("carrying capacity") — is a predictive software platform that helps Chilean municipalities anticipate the impacts of population change and infrastructure demand on the basic services they are responsible for delivering. Diario Sustentable described the system as "a semaphore that warns how much time remains before systems could collapse" — the platform forecasts five years ahead across ten core services and tells local decision-makers, in plain traffic-light terms, which ones they need to act on now.

CAPCA was developed at Pontificia Universidad Católica de Chile by an interdisciplinary team from the Instituto de Estudios Urbanos y Territoriales (FADEU) and the Facultad de Agronomía y Forestal, led by Prof. Roberto Moris. It grew out of the demographic-load study for Easter Island (Rapa Nui) that underpinned Chile's Law 21.070 on residency, and was funded by ANID through FONDEF-IT. Today the project is part of the HUBTEC tech-transfer portfolio and the Cero Labs UC initiative.

The scope of the work.

From a municipal user's perspective, CAPCA is a dashboard of ten core services — population, health, household waste, education, security, electricity, potable water, housing, motorization, formal tourism. Each service shows a traffic-light indicator against a latency band (the lead time a municipality actually needs to react — roughly two years, for example, to procure a new waste-collection truck) and a capacity ceiling. Behind each tile sits a research-grade projection model that forecasts five years ahead on the municipality's own historical data. A scenario builder lets planners drag three macroeconomic parameters — education rate, income rate, GDP — and watch the projected curves shift live; scenarios can be saved and exported as PDFs.

At the system level, CAPCA was built by a multidisciplinary team of eleven — frontend, backend, data science, data engineering, product, design — shipping a government-grade platform out of an academic research effort. The constraints stacked: simple enough for municipal staff with limited software background, robust enough to run reliably in production, and flexible enough to model communes as different as Rapa Nui (Easter Island), El Quisco, Pirque, and Renca — coastal, rural, dense, remote. And the system had to scale from the first pilot to the 11+ communes it runs in today, each with its own data ingestion and approval workflow.

Systems diagram titled EL QUISCO MUNICIPALITY showing how delivered services, operative division, directorate, community and national government interact, with failure modes labeled on the arrows between them.
The research diagram behind CAPCA: how a municipality's services buckle under demographic load — first modeled for El Quisco (whose population surges every summer) and then generalized to any Chilean commune.

My Role

Technical Lead · Principal Backend Engineer · Architect.

Hired to lead the backend, I ended up owning the full technical architecture of the platform — how the parts of the system fit together, how it ran in production, and how a multidisciplinary team of eleven shipped a government-grade tool together. The team brought deep expertise across disciplines: a senior frontend engineer, backend specialists, and a data-science core (including a PhD data engineer and a research analyst). My role was to shape the architecture that let all of that expertise land on a single operational platform.

Day-to-day, the work mixed three layers: hands-on backend development (the core API surface, the role-based permissions model, the data-upload and approval flows, the async job pipeline), architectural decisions across the frontend, backend, and data layers, and the engineering-practice work needed to move a university research project onto cloud infrastructure — containers, CI/CD, staging, monitoring. Toward the end of the engagement, a dedicated infrastructure engineer was brought in to take ownership of ongoing operations.

Key moments.

01

Designing for a municipal hierarchy

Chilean municipal work is strongly hierarchical — decisions flow through department heads, approvals happen at multiple levels, and nothing enters an official record without being signed off. That shape is baked into the platform: every commune has an administrator, every administrator invites department heads, every head has staff users, and every user has a role that controls what they can see and edit. Data uploaded by a junior user flows to their supervisor for review and approval before it ever enters the forecasting model. The hierarchy is not a bolt-on — it is the shape of the data model.

02

A UI for users who don't use spreadsheets

The primary users are municipal staff, not software users. Some had never opened a spreadsheet. That reshaped every UX decision: the dashboard is stripped down to traffic-light tiles with one number each; an in-app data-entry mode sits next to the spreadsheet upload so a user can type values into a grid without ever touching Excel; approval and feedback happen inside the app, with plain-language messages.

CAPCA main dashboard for a demo municipality: a grid of ten service tiles — population, household waste, health, education, security, electricity, water, housing, motorization, tourism — each with an actual value, a capacity ceiling, and a traffic-light indicator.
The CAPCA dashboard for a demo municipality: ten service tiles, each showing actual vs. capacity with a traffic-light indicator.

03

Bringing a research model into production

The forecasting engine at the heart of CAPCA is a research computation model that projects five years ahead from historical monthly data. It was written, naturally, in a tool better suited to research than to production services. Rather than rewrite it — which would have thrown away years of the research team's work — we wrapped it behind a clean service interface so the rest of the platform could call it like any other backend, and kept the research code where it belonged.

Expanded module view for household waste, showing three stacked charts — total, recyclable, and compostable — each overlaying historical monthly data in gray, a yellow latency band, and a red capacity ceiling over a 2021 to 2025 horizon.
Inside a module: the five-year projection of household waste against the latency band (yellow) and capacity ceiling (red), broken down into total, recyclable, and compostable streams.

04

Establishing engineering practice on a university project

A research team at a top university still runs, typically, on research tooling — local scripts, manual deploys, small test datasets. CAPCA needed to run as a government tool, for real municipalities, reliably. That meant introducing containers, CI/CD, cloud infrastructure, and an agile delivery cadence incrementally across the engagement — not as a mandate, but as practices adopted because they made shipped work easier. Late in the project, a dedicated infrastructure engineer was hired to take over ongoing operations.

Scenario builder view: a line chart on the left with seasonal historical data, a red maximum-capacity envelope and a yellow latency envelope across 2021 to 2025; a right-hand panel with an Escenario #1 name field and three sliders labeled Tasa de Educación, Tasa de Ingreso and Tasa PIB.
Scenario builder: adjust macroeconomic parameters on the right and watch the forecast curves shift live on the left; save and export as PDF.

The outcome.

  • In production in 11+ Chilean municipalities as of 2026 — five years after the first delivery. The system is still maintained and growing.
  • Pilots spanned Chile's geographic extremes: Easter Island (Rapa Nui) — the original demographic-load study behind Chile's Law 21.070 on residency — plus El Quisco (coastal, summer surge), Pirque (rural metropolitan fringe), and Renca (dense Santiago commune).
  • 10 core municipal services modeled per commune, each with a 5-year projection and a scenario builder on top.
  • Recognition trajectory: FONDEF-IT funding (ANID, 2019)Avonni 2023 finalistANID Startup Ciencia 2025 winner, with incorporation into the HUBTEC and Cero Labs UC tech-transfer portfolios.
  • Academic and business-press coverage including El Mercurio, Diario Sustentable, Diario Estrategia, Portal Innova, and UC's institutional channels (see press below).

Building software with a research team or a public-sector partner?

I work with groups that need someone who can own the architecture, write the backend, and set up production engineering — on projects where the goal is a live system, not a paper.

Let's talk