The opportunity
Klaviyo's AI & Analytics pillar is building the intelligence layer behind the product — models that optimize send times, predict churn, surface product recommendations, and power the agentic experiences that help marketers run their business autonomously. With over 200,000+…
What you'll do
Own the product strategy and roadmap for Klaviyo's core predictive and: generative ML models — including smart send time, audience optimization, product recommendations, and churn prediction — defining what "better" looks like and how we get there
Own the ML Platform roadmap: training infrastructure (DART/Ray), experiment tracking (MLflow), model serving, feature pipelines, and emerging tooling (Prefect) — making time-to-production for new models a first-class metric and driving it down continuously
Build and maintain evaluation and monitoring frameworks so model quality is: measurable, regressions are caught before they reach customers, and improvements compound over time
Partner with engineering leadership on build vs. buy decisions across the ML: stack — ensuring the platform evolves ahead of the needs of ML and AI product teams, not reactively behind them
Connect platform and modeling investments to customer and business outcomes,: working with go-to-market and customer success teams to ensure AI-powered features are well-understood and improving based on real feedback
Ensure platform reliability, observability, and cost efficiency for ML: workloads operating across hundreds of billions of events and 200,000+ customers
What they're looking for
- You have 5+ years of product management experience, ideally owning ML, AI, or: data platform products in a production environment — not just products that use AI as a feature
- You have working knowledge of ML systems: training pipelines, model serving, experiment tracking, feature stores, or related infrastructure — and understand the tradeoffs involved in building and operating them at scale
- You know how to connect internal platform investments to customer and: business outcomes, and can translate "we improved training throughput by 40%" into a product story leadership and go-to-market partners actually care about
- You're comfortable being the only PM in a highly technical room: you know when to drive decisions, when to defer to engineers, and how to build credibility without needing to be the most technical person there
- You think in terms of systems and tradeoffs: model quality, inference latency, training cost, developer velocity — and can make well-reasoned prioritization decisions when they conflict
- You can balance long-term platform investments with short-term product needs,: and know when to build, buy, or defer
- You communicate clearly and can translate complex ML and infrastructure: concepts into business impact for both technical and non-technical audiences
- You have a track record of moving roadmaps and priorities across ML, data: science, infrastructure, and product teams without direct authority
