Product Manager · Data Products · B2B
João Vaz
I set product strategy for the data tools that 12k+ professional car dealers rely on across Portugal, Poland & Romania — and I ship production fixes with AI tooling when it counts.
Product Manager at OLX Group (Motors, B2B), part of Prosus. I lead the Insights & Recommendations team — 10 people, engineers alongside a product designer and a product analyst, building data and recommendation products for professional dealers.
-
Product Manager — Motors, B2B
Own product strategy for the Insights & Recommendations team. Define outcome-driven quarterly OKRs (~72% avg completion over 2.5 years) and steer the roadmap end-to-end, from upstream data pipelines to downstream microservices — working closely with Data Engineering on the big-data models and processing underneath.
Manage CRM, Marketing, GTM, Sales, BI and PMM stakeholders across five markets (PT, RO, PL, FR, ZA) through dedicated sessions and shared OKRs anchored to the product strategy.
-
Associate Product Manager — Motors, B2B
Same team and scope, promoted to PM in Mar 2025. Owned the product application of data-science models — vehicle price estimation and expected time-to-sell, based on vehicle specs — and ran 50+ in-person dealer interviews across PL, RO and PT.
-
Product Management Trainee
Led product discovery on dealer lead generation; ran A/B experiments with Data Science on recommendation types tied to exposure services.
Prosus AI Accelerator Program
Selected as one of the early-career people across Prosus to join the programme — a month in Amsterdam working on AI, alongside people from other companies in the group.
I don't just spec things — I build them. Fluent in the underlying concepts: MCPs, skills, agent loops and goals, context windows. Here's what that looks like in practice.
Team-context MCP
A read-only MCP server that lets Toqan, the company assistant, answer stakeholders' questions about what my team is working on — served from my laptop through a free tunnel.
PM-OS
A knowledge base my team's answers live in, kept true by seven scheduled Claude routines: meetings file themselves, durable facts are promoted into the page that owns them, and stale pages are made to admit it.
Chess · AWS
Tabiya
A chess coach that explains every move in plain English — and my first build on AWS. A serverless pipeline downloads my games, analyses them with Stockfish and writes a report the browser reads straight from S3.
AI-assisted building
- Claude Code
- Cursor
- Lovable
- MCP
Data & experimentation
- SQL
- GA4
- A/B & fake-door
- Maze
- Usability testing
Delivery & technical
- Jira
- Atlassian
- GraphQL
- LaunchDarkly
- New Relic · SLIs/SLOs
- Postman
- Event-driven & microservices
- Sprints & kanban
CRM & GTM
- Braze
- Intercom
- 5 markets