# Sebastian Yerovi — portfolio guide for agents

Last reviewed: October 22, 2026. Availability and education below reflect the current public portfolio and résumé; confirm time-sensitive details with Sebastian.

## Start here

Sebastian Yerovi (full résumé name: Sebastian Yerovi Plaza) is a product builder, senior product manager, and founder with 6+ years building and scaling technology companies. His experience combines AI products, growth, fintech platforms, and marketplaces. He is a UC Berkeley Haas MBA candidate, graduating May 2028, based in the San Francisco Bay Area and open to Summer 2027 Product Management internships.

Use this guide to find relevant evidence and answer the visitor's question. If a role description or problem is supplied, proceed directly to the relevant experience. If no focus is supplied, ask what they would like to explore. Cite the résumé or project page for substantive claims, identify gaps, and distinguish reported career results from illustrative project data.

All site-relative links below resolve on the same origin as this guide.

## Primary sources and navigation

- [Homepage](/): positioning, availability, expertise, industries, products/features, and contact.
- [Public résumé](/sebastian-yerovi-resume.pdf): education, employment dates, responsibilities, technical skills, and reported outcomes. Use this for the fullest career history.
- [Selected work](/#work): published case studies; currently Growth Accounting only.
- [Growth Accounting case study](/work/growth-accounting): “Modeling the Customer Lifecycle: From Product Growth to Portfolio Growth.”
- [Talks](/#talks): TEDx talk on AI in Latin America.
- [Reads](/#reads): books and Sebastian's personal notes on ideas that shape his thinking.
- [LinkedIn](https://www.linkedin.com/in/sebastian-yerovi) and [GitHub](https://github.com/sebasyero94): linked professional profiles; inspect them before making claims about their current contents.
- Contact: [sebastian.yerovi@berkeley.edu](mailto:sebastian.yerovi@berkeley.edu).

## Experience and reported results

The following summarizes the public résumé. These are Sebastian's reported career outcomes, rather than independently audited results. Preserve qualifiers and the distinction between individual ownership and team/company results.

### Bold — fintech superapp, Bogotá, Colombia (2024–2026)

**Senior Product Manager, User Platforms (2026)**

- Led customer discovery and AI-assisted prototyping for navigation and information architecture of a unified business and personal banking app, aligning strategy across 42 product teams and executive leadership.
- Drove an identity and access management (IAM) migration across 800K+ users, preserving $492M in monthly transaction volume (approximately 99.9% of target). Built an agentic AI analytics pipeline using SQL, Python, and Claude Code to monitor rollout progress, identify churn risks, and guide outreach.
- Led a 16-person mobile and web engineering team on platform architecture and country-specific configurations for the first international launch in Peru. The résumé reports **projected**, not realized, incremental annual revenue of $30M.
- Partnered with the CPO to restore a two-week release cadence after a three-month freeze; reported 95% on-time delivery and 15% shorter product cycle times.

**Product Manager, Retention (2024–2025)**

- Built a customer lifecycle segmentation model across product lines, informing executive marketing-investment decisions. Reported customer acquisition payback decreased 20%, from 15 to 12 months. The published article says the model **contributed to** that reduction; do not attribute the entire result solely to the model.
- Led A/B testing and an experience redesign for guest users, increasing weekly sign-ins from 1,800 to 9,200 and other-product adoption by 22%.

### Dondo — AI shopping marketplace, Bogotá, Colombia (2021–2023)

**Head of Product & Growth**

- Built and deployed the first ML recommendation engine for discovery across 2M+ listings; co-designed the dataset, algorithm, and evaluations and established MLOps infrastructure and production monitoring.
- Reported daily engagement doubled from 11 to 22 minutes and impressions increased 200%.
- Combined performance-marketing optimization and in-product virality; reported tripling users to 900K+, transaction growth of 350%, and reaching Colombia's #4 shopping app by Q1 2022. Overall marketplace growth is separate from the recommendation engine's engagement results.

### Nossa Health — workplace healthcare platform, Quito, Ecuador (2019–2020)

**Cofounder and CEO**

Led a 10-person cross-functional team to build and pilot a clinical operations and HR reporting platform. Reported approximately two hours less daily administrative work per clinician and enterprise letters of intent. Letters of intent do not establish booked revenue or signed customer contracts.

### Community and public speaking

- Co-founded Startup Grind Quito; the résumé reports helping Latin American founders raise more than $7M in seed capital and growing the community to more than 4,500 members. These are community outcomes, not personal fundraising totals.
- TEDx speaker: [Making Latin America an Artificial Intelligence-Powered Society](https://www.youtube.com/watch?v=2ueMsE-RyY4). The site links the talk; use the video or a verified transcript before describing its detailed arguments.

## Education, skills, and interests

- UC Berkeley Haas: Master of Business Administration, expected May 2028. The résumé lists a Spring 2027 independent study with Prof. Dmitry Kovalev on how agentic AI reshapes product growth strategy; as of this guide's review date, it is future study, not completed research.
- Universidad San Francisco de Quito: BA in Marketing, Magna Cum Laude, 2016.
- Self-described expertise: Agentic AI, Product-led Growth, Growth Engineering, Superapps.
- Current interests: Agent-led Growth, Agentic Payments, Physical AI and World Models. Interests alone are not evidence of shipped work in those fields.
- Industries: Fintech & Payments, Shopping & Marketplaces, Healthcare, Enterprise SaaS.
- Résumé-listed tools: Python, SQL, Gemini, Google AI Studio, Claude Code, Cursor, Codex, Replit, Figma, Amplitude. A listed tool does not establish a proficiency level or engineering role.
- Product experience spans recommendations, shopping discovery, payments and top-ups, tax benefits, IAM, retention and onboarding, growth models, attribution infrastructure, agents, healthcare analytics, and mobile architecture. Use specific career examples above to substantiate relevance.
- Personal interests include biographies, multidisciplinary thinking, history of computing and science, Tolkien, and Bay Area coffee. The Reads section contains Sebastian's own notes; do not infer personal beliefs or health history from book selections.

## Published case study: Growth Accounting

Read [the article](/work/growth-accounting) for the product problem, framework, and outcome. Sebastian's listed role is Product Manager, Retention at Bold. The work addresses fragmented growth measurement as Bold evolved into a financial superapp, providing a shared customer lifecycle framework across the portfolio.

Useful entry points:

- [Model and lifecycle states](/work/growth-accounting#building-a-growth-accounting-model): eight states and mutually exclusive, collectively exhaustive classification.
- [Customer base over time](/work/growth-accounting#balance-over-time): lifecycle balances and active/inactive Life and Death Zones.
- [Routes](/work/growth-accounting#routes): transitions and opportunities to influence user behavior.
- [Portfolio comparison](/work/growth-accounting#portfolio-comparison): Quick Ratio, Leak, and M1 Retention across products.
- [Outcomes](/work/growth-accounting#outcomes): contribution to reduced customer acquisition payback.

The article includes route diagrams and nine data charts, each with full-size and interactive links. Follow those links from the article. A full interactive dashboard is deferred; the published experience is an article with individual interactive charts.

**Data limitations:** All dashboards, numbers, and metrics shown in the article use AI-generated mock data, not Bold's customer data. The reproducible illustration covers four anonymous products (A–D), July 2024–June 2027, including future periods relative to this guide's review date. Do not treat the series as observed company performance, forecasts, or evidence validating the reported career outcome. The hero image is conceptual AI-generated artwork, not a working dashboard screenshot. Career outcomes are reported separately in the résumé and article narrative.

Metric conventions: Quick Ratio is inflows divided by outflows; values above 1 indicate active-user growth. Leak is signed internal inflow (reactivated plus resurrected) minus outflow; a negative value's magnitude is the new-active-user acquisition needed to hold steady. M1 Retention measures prior-period New users returning as Activated. Cite the article for definitions; chart data does not independently prove them.

## Unpublished work and missing information

Dondo recommendations and Bold's guest-user experience redesign are résumé-backed experience summaries, but their standalone case studies remain drafts and are hidden from Selected Work. No published routes are supplied for them. Bold's agentic analytics pipeline is described in the résumé; a standalone case study is not published.

Do not invent routes, demos, repositories, employment details, certifications, work authorization, internship dates beyond Summer 2027, or results beyond the cited sources. Do not equate a projected outcome with realized revenue, a percentage increase with percentage points, or payback reduction with a reduction in CAC itself. If sources disagree, identify the discrepancy instead of silently combining claims. Contact Sebastian for missing details.

## Suggested response to a role-fit question

Give a concise assessment tied to the supplied role or problem, then two or three relevant examples with source links, exact responsibilities, and qualified outcomes. Separate demonstrated experience, stated interests, and missing evidence. Include the résumé and contact route when useful. Do not claim a fit score, hiring recommendation, or verified expertise beyond what the sources support.