Build an AI Growth Optimization System
.png&w=1536&q=75)
Learn directly from Eric Metelka
What you'll learn
Identify where evals end and experiments begin
Recognize where evals stop giving signal and when to run a live experiment
Layer evals, experiments, and analytics
Apply a practical framework for sequencing eval checks, online A/B tests, and Amplitude analytics across the AI product
Automate your experimentation loop
Use AI-native tooling to generate hypotheses, run tests, and surface analysis without manual overhead at each step.
Why this topic matters
Evals tell you your model is working. They don't tell you it's working for your users. The jump from offline evals to live experiments is where most AI product teams stall, and where the teams that don't pull ahead. This session shows you how to build the full loop.
You'll learn from
Eric Metelka
Product Leader, Amplitude Experiment | Lenny's Top 25 Contributor
Eric Metelka leads product for Amplitude Experiment, where his team builds the experimentation and feature flag platform. He has 11 years of experience building growth and experimentation systems across marketplaces, SaaS, and consumer platforms, and is a top 25 contributor to Lenny's Newsletter, where he advises PMs and growth professionals on product and career strategy.
Connect with Eric in Cohort, an Amplitude community for product leaders and AI builders.
