Senior Product Manager · Head of Product

The problem underneath the problem.

Most teams are solving the wrong problem, efficiently. I identify the real constraint, build the evidence to act on it, and design the intervention that holds up after I leave. 15+ years across healthtech, proptech, and public sector digital transformation.

See the work →
5 days 2 min

Tenant evaluation time. RentScore, TangoRent by Greystar, a Fortune 500 proptech platform rebuilt around government data APIs instead of manual review.

6.2 1.3 days

Cardiac average stay. HSJD Emergency Redesign, recognized at Stanford's Lean Healthcare Conference, 2017.

94.95%

Model accuracy, 1,487 patient records. Never implemented, the most instructive result in this whole portfolio.

Case Studies
Proptech · TangoRent by Greystar (Fortune 500) · 2023

RentScore: tenant screening from 5 days to 2 minutes

Manual tenant evaluation was the bottleneck killing contract velocity. I redesigned the process around government data APIs and automated scoring, eliminating fraud and friction simultaneously.

Read the case study →
5 days → 2 min
Evaluation time
2–4 → 8–10
Contracts / week
72 hrs → 0
Signing delay
AI Product · Claude + Claude Code + Vercel · 2026

Revenue Leak Calculator: full PM process on an AI-native build

Healthcare clinics lose revenue to invisible process failures, but can't quantify it. I applied full product discipline: JTBD, PRD, architecture spec, data strategy, and shipped a live product in 4 days using AI as the development lever.

Read the case study →
4 days
Problem to deployed
Live
calculadora.dentalgrowstack.com
Healthtech SaaS · Next.js + Supabase · 2026BETA

Kira: emotional support between dental appointments

High-ticket dental clinics lose patients mid-treatment to anxiety, not clinical failure. Kira tracks patient emotional state between visits and alerts clinics before someone disappears. Built with full PRD, design system, and technical architecture before the first line of code.

Read the case study →
4.2/5
Viability score
8 stories
All P0 · pilot-ready
Pilot
Entering validation
Healthtech · Universidad de Chile / HSJD · 2016–2017

Cardiac Emergency Redesign: from 6.2 to 1.3 days

The emergency department was drowning in overcrowding. I mapped patient flows from security camera footage, identified systemic coordination failures, and redesigned the service, down to the architectural floor plan.

Read the case study →
6.2 → 1.3 days
Cardiac avg stay
−49%
Total wait times
Stanford
Lean Healthcare 2017
Clinical Analytics · Clínica Las Condes · 2010–2011

Predictive Risk Stratification for Chronic Disease Management

Type 2 diabetes patients weren't being stratified by cardiovascular risk: every patient received the same attention regardless of risk level. I co-designed the clinical criteria with endocrinologists, built the model from 1,487 paper records, and architected the full process. 94.95% accuracy. Never implemented, and that's the most instructive part.

Read the case study →
94.95%
Model accuracy
1,487
Patient records
0.972
Precision · high-risk class
The Lab

I use AI across the full PM workflow, not just for writing: discovery synthesis, assumption pressure-testing, rapid prototyping with Claude Code, spec drafting. JTBD, strategic framing, and architecture decisions are product judgment calls; Claude executes them. These are products I build myself, with AI as the development lever. Not just advised, shipped.

RūfCheck / Adjudica

AI for public health tenders in Chile. Catches compliance errors before they cost the bid.

Claude Code · n8n · Vercel
In development, Sprint 6
DentalGrow / Smile360™

AI-driven growth agency for high-ticket dental clinics.

GHL · n8n · Meta Ads
Active
Revenue Leak Calculator

Interactive revenue-leak calculator as a lead magnet. Full PM process, shipped in 4 days.

Vercel · Supabase
Live
Methodology

The pattern across every project

Every case study in this portfolio follows the same underlying structure, not because I applied a template, but because complex problems tend to hide in the same places.

01 · Diagnosis
The problem is rarely where it appears

In RentScore, everyone said the problem was speed. It was fraud architecture. In HSJD, the complaint was overcrowding. The actual failure was in the coordination layer between units. I don't start with the stated problem. I start with what the data and the people closest to the work actually show.

02 · Discovery
Jobs-to-be-Done + Opportunity Solution Tree

I use JTBD and OST (Teresa Torres) as thinking tools to resist the pull toward solutions before the problem is understood. Weekly user contact. Assumption mapping. The riskiest assumptions tested first, cheaply, before any engineering investment. Discovery doesn't end when you start building. It's when the real signal starts.

03 · Measure
The metric and its guardrail, defined before the spec

Every project in this portfolio has a North Star metric and an explicit counter-metric defined before the solution was designed. If I can't name what would tell me the solution is working and what would tell me it's gaming the metric, I haven't thought hard enough.

The constraint that shows up everywhere

The hardest constraint in complex service environments is almost never technical. It's organizational readiness: the gap between what the evidence supports and what the institution is structured to adopt. I've learned this in a public hospital, a private clinic, a Fortune 500 proptech platform, and my own products. The Clínica Las Condes case study documents this most explicitly: 94.95% model accuracy, never implemented, and that's the most instructive outcome of the five.

What this looks like in practice
  • Go to the floor before the whiteboard
  • Name the counter-metric before the spec
  • Identify the internal champion, not just the sponsor
  • Document what didn't work and why
About

I'm strongest when the problem isn't defined yet.

Not “we need a feature.” It's “something is broken and we don't know what.” That's where I start: figuring out what's actually broken before designing anything.

My core loop: diagnose what's real, redesign how it works, automate so it doesn't break again.

That loop has played out in a national public hospital during a winter emergency peak, inside a Fortune 500 proptech platform scaling across Latin America, at Chile's Ministry of Health designing surgical patient flows for an entire country, and in products I've built myself with AI as the development lever.

The domain changes. The thinking doesn't.

Based in Santiago, Chile (GMT−4). Available for remote and hybrid roles globally. Fluent in English and Spanish.

Erika Quiroz
  • MIT Sloan Executive Education
    Organizational Design for Digital Transformation
  • MBE · Universidad de Chile
    Master's in Business Engineering with IT
  • Stanford Lean Healthcare Conference
    Only Latin American work presented · 2017
  • Serie Sistemas Públicos N°14
    Co-authored with Prof. Óscar Barros, U. Chile
  • Anthropic Certifications
    Claude Code · Claude Platform · Code in Action · Agents
Contact

Open to Senior PM roles, fractional engagements, and consulting.

Prefer LinkedIn?LinkedInGitHub