Hannah Cepik

AI Product Management · Growth & Experimentation

Products that
learn faster
than the market

I'm an AI product manager with 5+ years across B2B marketplaces, SaaS and media. I build the growth machinery itself — analytics infrastructure, experimentation frameworks and AI-powered distribution systems — then use it to ship, measure and scale what works.

Rio de Janeiro, BR Remote — US & EU teams EN / PT bilingual
2× conversion targets — Better Collective PRD → live product in <20 days — Abyss AI $39.8B GMV platform — BEES / AB InBev Time-to-value cut from 14 days to 5 Short-form content at near-zero marginal cost 2× conversion targets — Better Collective PRD → live product in <20 days — Abyss AI $39.8B GMV platform — BEES / AB InBev Time-to-value cut from 14 days to 5 Short-form content at near-zero marginal cost

Selected work

3 case studies

Abyss AI · Animation streaming · 2026

Fan Universes

A UGC engine that lets fans create their own AI-animated scenes from Aby shows — from PRD to live product in under 20 days.

Aby is a first-of-its-kind AI animation streaming platform. The bet behind Fan Universes: fans don't just want to watch a universe, they want to make things inside it. I owned the product end to end — wrote the PRD, built an interactive prototype to align engineering and leadership fast, defined the event taxonomy in PostHog before launch (not after), and wired Klaviyo lifecycle flows so every creation moment fed retention.

Prototype-driven development compressed the usual discovery-to-delivery cycle: two show launches shipped on the engine within the first sprint window.

<20 daysPRD to live product
2 showslaunched on the engine
Day onefull-funnel analytics
Product ManagementAI Product StrategyPostHogKlaviyoPrototype-driven dev
aby.one ↗

Better Collective · Sports media · 2024–25

Experimentation Program

Built the A/B testing framework from scratch that doubled conversion targets across high-traffic US sports media assets.

The product had traffic; what it lacked was a disciplined way to learn from it. I established the full experimentation practice — hypothesis standards, prioritization, statistical guardrails, and a testing cadence the whole team could run on. I wrote and shipped A/B tests in JavaScript myself, closing the gap between "idea" and "live experiment" from weeks to days.

The framework became the engine for revenue acceleration: conversion targets were not just met but doubled, and voice-of-customer interviews fed a prioritized backlog that kept engineering pointed at validated pain, not opinions.

2×conversion targets hit
0 → 1experimentation practice
Weeks → daysidea to live test
Growth ExperimentationA/B TestingJavaScriptGTM & PricingVoice of Customer

Abyss AI · Distribution & automation · 2026

Faceless Content Automation Pipeline

An automated short-form video system — Claude → Veo3 → Python → n8n — producing high-volume content at near-zero marginal cost.

Growth on TikTok is a volume game, and volume is usually a headcount problem. I turned it into a systems problem instead: Claude generates scripts and hooks, Veo3 renders the video, Python assembles, and n8n orchestrates publishing — a pipeline that ships content while the team sleeps.

Automation without governance is just noise, so every content line runs on a 14-day kill/scale framework: UTM-level attribution decides what earns more budget and what dies. AI here isn't a productivity gadget — it's a core component of the growth stack.

~$0marginal cost per video
14-daykill / scale framework
UTM-levelattribution on every post
AI Agents & Automationn8nClaudeVeo3PythonGrowth Experimentation

Experience

2021 — today
2026 — now

AI Product Growth Strategist · Abyss AI, New York (remote)

Own the full growth stack for an AI animation streaming platform: analytics infrastructure, attribution pipelines, channel experimentation across five platforms, creator partnership ops, and an AI-powered marketing intelligence hub delivering a daily executive briefing.

2025 — 2026

AI Product Manager · Abyss AI

Owned discovery and roadmap for the early-stage platform; shipped features 25% faster with AI-powered prototypes and laid the analytics foundations (event taxonomy, funnels) the growth role later scaled.

2024 — 2025

Product Manager, Growth & Experimentation · Better Collective, US/Denmark

Built the experimentation framework for a high-traffic sports media product; led AI/ML feature launches and the go-to-market strategy — value proposition, pricing, and customer-facing messaging — that lifted conversions.

2022 — 2024

Product Manager & UX Research · BEES (AB InBev)

Product strategy on a $39.8B GMV beverage distribution platform: cut time-to-value from 14 days to 5, led research across 6 international markets (+35% segment engagement), managed P&L and a cross-functional team of 10.

2022

Product Manager · Purple Metrics

Cut market-fit validation from 4 weeks to 10 days with voice-of-customer practice; PRD standards adopted company-wide.

2021 — 2022

Product Manager · Runrun.it

Built a work management product generating R$3M+ in contract value, including the form-to-task automation that became the company's strongest enterprise differentiator.

2013 — 2020

Researcher · Federal University of Minas Gerais

Led 50+ academic research projects in anthropology and social sciences, including field studies on human-computer interaction — the research rigor behind everything above.

Toolkit

used in production, not on slides

Experimentation

  • A/B testing (VWO, JS)
  • Funnel diagnostics
  • Kill / scale frameworks
  • Pricing & GTM tests

Analytics

  • PostHog
  • GA4 · Mixpanel
  • SQL
  • UTM & attribution design

AI & Automation

  • Claude (agents & workflows)
  • n8n orchestration
  • Veo3 · Python pipelines
  • AI-powered prototyping

Product craft

  • PRDs & engineering specs
  • User research (6 markets)
  • Roadmaps & North Star metrics
  • Cross-functional leadership

About

the person behind the funnels

Before product, I spent six years as a social science researcher at one of Brazil's top universities — anthropology, field studies, quantitative analysis. That training is my unfair advantage: I treat every product like a field site, every metric like an interview that hasn't been asked the right follow-up question yet.

Today I blend that research rigor with hands-on execution. I write the PRD, prototype it, instrument it, ship it, and read the data myself. I'm most at home in the space where AI stops being a productivity tool and becomes part of the product's growth engine.

Based in Rio de Janeiro, fully bilingual, and experienced working remotely with teams in the US and Europe.

BaseRio de Janeiro, Brazil
LanguagesPortuguese · English (bilingual)
EducationBSc Social Sciences (UFMG) · MSc Social Sciences (UFRRJ)
FocusAI products · Growth · Experimentation
CertifiedAI Product Leaders · Digital Product Leadership