Felipe Stenzel
Product leader and AI‑powered builder with 10+ years in K–12 at Latin America's largest education group. I design, code, and ship software end‑to‑end, and I form teams that do the same.
Impact Metrics
Eight years helping build Brazil's leading K–12 platform, from ~100K to 6M+ students.
Complete software solutions: backend, frontend, DevOps — plus 10+ landing pages. Built solo, end-to-end.
From PM to builder: leading product strategy, venture creation, and hands-on development at scale.
About me
I lead innovation and technology at Cogna Labs, the innovation arm of Latin America's largest education group. Labs works like an internal startup incubator — we build ventures from scratch, leading cross‑functional teams to turn ideas into products and businesses.
As a naturally curious person, I was always frustrated as a PM by not being the one actually building. Today, with AI as my copilot, I can ideate and build complete software end‑to‑end. I've launched three major internal products, two built entirely solo — combining strategy, design, and code.
My mission: form AI‑powered builder teams — small, multidisciplinary squads where each person masters their craft but also understands the full product lifecycle. Smaller, faster, sharper.
Toolkit — Tech, Product & Business
What I bring to the table: technical execution + product thinking + business acumen.
Built solo, then scaled with teams
From frustrated PM to full-stack builder. These are complete software products I designed, architected, coded, and shipped to production — entirely solo. Backend, frontend, DevOps, data engineering. When AI became my copilot, I stopped asking engineers to build my vision. I built it myself. Then, as these projects proved their value and gained traction, they were incorporated into the company's tech structure — with proper handover to engineering teams for scalability, maintenance, and continued evolution. I don't just prototype; I build production-ready systems that become foundational infrastructure.
B2G Data Lake & Intelligence Platform
Government decisions on education, procurement, and public policy were being made with fragmented data across hundreds of sources. This platform unified the chaos: a nationwide data lake covering all 27 Brazilian states + 5,570 municipalities, ingesting public datasets (IBGE, TSE, Siconfi), partner data, and operational systems into a single source of truth for strategic decision-making.
🎯 Impact — From scattered spreadsheets to real-time intelligence. Used by secretariats, internal teams, and executives for cross-dataset analysis (elections × education × finance) that was previously impossible.
⚙️ Architecture & scale
All 27 states + Federal District — complete national coverage
ELT/ETL with batch & streaming ingestion, data contracts, validation, observability
IBGE (census/geo), TSE (elections), Siconfi (finance), education indicators, partner integrations
🚀 Key capabilities
Single source of truth covering every Brazilian municipality. Raw → refined → semantic layers with partitioning, SCDs, and access policies.
Batch & streaming loaders for government APIs, partner systems, and operational data. Schema registry, data contracts, validation, and full observability.
Dashboards for secretariats, anomaly alerts, public APIs. Unique analyses combining elections × education × finance data for strategic insights.
Entity resolution (municipalities, schools), trend detection, natural-language queries. Secure "exam-mode" for high-stakes evaluations.
🏗️ My role & ownership
Designed the complete data model, governance framework, and technical architecture for nationwide coverage (all 27 states + 5,570 municipalities).
Built ingestion & transformation pipelines, standardized data contracts for partner integrations, implemented schema registry and validation layers.
Created operational dashboards and APIs for internal teams, external stakeholders, and secretariats. Delivered cross-dataset analyses that directly informed strategic decisions at board level.
Live State Budget Ranking
LIVE — Siconfi APILive technical demo: Real-time consumption of Brazil's National Treasury public API. 27 parallel requests fetching state budget data — each response contains ~150 nested fields across multiple report types. Frontend filters relevant entries, extracts target metrics, aggregates values, maintains dynamic sorted rankings, and renders progressive UI updates as data streams in. This demonstrates the core data engineering patterns (API orchestration, JSON parsing, filtering, aggregation, real-time sorting) that power the full Data Lake platform — which operates at significantly larger scale with dozens of data sources, millions of records, and complex cross-dataset joins.
Fetching data from Siconfi API...
⚙️ Technical architecture: Node.js proxy handles CORS + API gateway logic. Frontend orchestrates 27 concurrent async requests, receives ~4KB JSON per state (total ~108KB raw payload), applies multi-level filtering (report type → column → account code), extracts numeric values from nested objects, performs in-memory aggregation, maintains sorted state with O(n log n) re-ranking on each update, and renders progressive UI with sub-200ms latency per state. All data transformations (parsing, filtering, calculations, formatting) run client-side. This mini-demo showcases my full-stack capabilities (backend proxy, frontend data orchestration, real-time UI) — the actual Data Lake B2G product is orders of magnitude more complex, with distributed ingestion pipelines, data warehousing, semantic layers, and multi-tenant analytics APIs.
Control Tower — Venture Builder Management Platform
"Jira didn't work for us — so I built our own." Managing multiple ventures, financial workflows, and cross-functional teams with off-the-shelf tools became a bottleneck. Control Tower is a full-stack enterprise platform designed for venture builders: combines project management, financial operations, and team coordination into a single source of truth. Handles everything from discovery to scale, with built-in budgeting, approval workflows, and real-time collaboration.
🚀 From frustration to production — Built solo as a complete replacement for Jira/Trello/Asana. Now powering venture operations with 18 database tables, 37 API endpoints, and enterprise-grade security (RBAC, audit logging, rate limiting).
⚡ Technical stack & architecture
PostgreSQL (Neon) with Prisma ORM — Users, Ventures, Tasks, Budgets, Requests, Audit Logs
Next.js API routes with Zod validation, rate limiting, and full RBAC enforcement
Admin, Manager, Venture Creator, Financial Analyst, Marketing — granular permissions
🎯 Key features
Track ventures through 6 phases (Discovery → Scale), 3 organizational modes, budget tracking, phase history, and 14 market/product fit categories.
5 request types (purchases, hires, reimbursements, subscriptions), approval workflows, CSV import, budget forecasting, multi-currency support, contract tracking.
Flexible structure (Stages → Epics → Tasks → Subtasks), Kanban boards, 4 priority levels, 5 status types, delay indicators, commenting system.
Venture analysis, financial reports, budget vs actual, phase duration metrics, custom KPIs, export to XLSX/CSV.
🔐 Security & architecture
Clean separation: domain logic, use cases, infrastructure adapters. Repository pattern, dependency injection, middleware chains.
NextAuth JWT sessions, bcrypt hashing, rate limiting (10 login attempts/15min), SQL injection protection (Prisma), audit logging (CRUD + auth events), strict CSP headers.
Deployed on Vercel with SSL, error boundaries, loading states, optimistic updates, React Query caching, TypeScript strict mode, Jest testing setup.
Interactive Kanban Board
Drag cards between columns to see real-time updates. This is the core interaction pattern that powers Control Tower's task management.
💡 Why I built this: Commercial tools like Jira are bloated with features we don't use and lack the specific workflows venture builders need. Control Tower is purpose-built for our operations: venture lifecycle tracking, financial approval chains, and hierarchical task management — all in one place. The result: faster decisions, clearer visibility, and better resource allocation.
RADAR — Startup & Market Intelligence
Finding the right startups to partner with, invest in, or compete against used to take weeks of manual research. RADAR turns that into minutes: a strategic intelligence system tracking ~3M startups (Crunchbase-scale), mapping them to company objectives, and deploying AI agents that surface insights, score strategic fit, and generate executive briefings automatically.
🏗️ Built 100% solo — from database schema to production deployment. Full-stack architecture handling 27.2M records across 22 tables, serving real-time insights to cross-functional teams.
⚡ Technical stack & scale
22 tables, 142 metadata fields, Neon PostgreSQL
FastAPI backend (Auth, Analytics, Pipeline, Startups, Radar) + React/Next.js frontend
Rendered instantly on Visual Radar screen with real-time filters
🎯 Key capabilities
Real-time translation, smart filters, instant results across millions of companies. Find exactly what you're looking for in seconds, not hours.
LLM agents analyze startups against strategic criteria, score fit, flag risks, and generate executive briefings — automatically.
80M+ companies rendered on interactive maps with heat visualization. Dashboards with geo analytics, trend charts, and KPI tracking.
Kanban-style opportunity tracking with CRUD operations, status updates, and team collaboration features.
Agents in action (simulated)
Prompt
"Find startups in adaptive learning aligning to 'K–12 outcomes' and 'low‑TCO deployments'."
Plan
Index → filter by tags → score by strategy fit → surface top 5 with risks.
Output
[1] Company A — high fit (K‑12, infra‑light)…
Scaling products at enterprise level
Alongside building products solo, I spent 8 years as PM and Group Product Manager at Brazil's largest EdTech, leading 10+ different squads and scaling a platform from 100K to 6M+ active learners. This dual experience — leading teams at enterprise scale AND building hands-on — gives me a unique perspective on both strategic product leadership and technical execution.
Plurall — K–12 Learning Platform
Brazil's leading K–12 platform serving schools, states (B2G), teachers, students and families — from early childhood to high school and pre-vestibular. A true super-app for education: assessments, adaptive learning, teacher tools, parent engagement, content distribution, and B2G integrations.
🎯 8 years, 10+ squads, 6M+ users — Rose from Product Owner (2016) to Senior PM to Group Product Manager, leading cross-functional teams (Design, Engineering, Data, QA) across multiple product areas simultaneously. Shipped features serving millions while maintaining 99.9% uptime during national exams.
💼 Leadership & impact
Led cross-functional squads across Assessments, Adaptive Learning, Content, Teacher Tools, B2G integrations, Parent Platform, Mobile Apps, and Growth — managing teams of up to 15+ people (Design, Engineering, Data, QA, Analytics).
Defined quarterly OKRs, prioritized features based on user research + data, balanced business goals (revenue, retention, NPS) with technical debt and platform stability. Ran discovery, user testing, A/B experiments.
Partnered with Sales, Marketing, Customer Success, Pedagogy, State governments (B2G contracts), school directors, and C-suite. Presented roadmaps to board, negotiated scope with state secretariats.
🚀 Key achievements
Platform grew 60x while maintaining quality. Shipped features for B2C (students/parents), B2B (schools), and B2G (state governments). Led migrations, redesigns, and architecture modernizations.
White-label assessments & adaptive study flows for 10+ teaching systems. Built personalized learning paths, automatic remediation, video-supported review, secure exam modes, real-time teacher dashboards.
Shipped platform features serving entire state education networks. Handled procurement processes, compliance requirements, data privacy (LGPD), and massive concurrent usage during synchronized state-wide exams.
Mentored junior PMs, established discovery processes, introduced data-driven decision-making, created PRD templates, ran retrospectives, implemented continuous delivery. Reduced release cycle from monthly to weekly.
📚 What I learned
6M users means every bug affects thousands. Every decision requires stakeholder alignment. Every release needs monitoring, rollback plans, comms strategy. I learned to balance speed with reliability.
As PM/GPM, I spent significant time in meetings, aligning teams, writing specs, negotiating priorities across multiple stakeholders. This taught me how essential clear communication and structured processes are when operating at enterprise scale.
Parallel to my PM work, I've always been naturally drawn to building directly — understanding systems from the inside. With AI as my copilot, I can now design, code, and ship complete products solo for rapid 0→1 validation, then hand off to engineering teams for scale. My Plurall experience taught me what users need at scale; my builder projects prove I can execute end-to-end when speed matters.
Impact Timeline
💡 The complete picture: Plurall taught me how to lead, prioritize, and ship at scale. My builder projects show I can also execute hands-on when needed. Together, this gives me the rare ability to operate as both strategic product leader and hands-on technical contributor — switching modes based on what the challenge requires.
Career timeline
Currently at Cogna Educação (since 2016) — Brazil's largest education group — where I've grown from Product Owner to Innovation & Technology Coordinator (Staff level). Also teaching Product Management at Tera (2022–present), training the next generation of PMs.
Leading innovation arm: building ventures from scratch, coordinating cross-functional teams, shipping 0→1 products (RADAR, Control Tower, Data Lake B2G).
Built internal startups end-to-end: from ideation to production deployment, combining strategy, design, and full-stack development.
Teaching Product Management fundamentals: discovery, roadmapping, metrics, stakeholder management. Mentoring aspiring PMs.
Led multiple product squads simultaneously. Defined strategy, prioritized roadmaps, managed cross-functional stakeholders.
Scaled Plurall from 100K to 6M+ users. Led Estudo Orientado (adaptive learning), assessments, B2G expansion.
First PM hire. Established product processes, discovery rituals, and data-driven culture. Started with one squad.
Digital marketing strategy, analytics, and growth optimization for enterprise clients.