skills and experience with unstructured data. You'll build prediction and optimization models across the business, and help us expand into new territory: extracting signals from user prompts, photos, and video to power ad prediction, user intent modeling, recommendations, and personalization. You'll also contribute to the team's AI ecosystem, including evaluating and calibrating AI agent outputs. Our stack: Python, BigQuery, GCP, our own production pipeline framework (Nova), Prefect, and Grafana. We're looking for someone with strong modeling instincts who cares about business impact — not just technical depth, but making sure the work lands and changes how decisions are made. What You Will Be Doing Build prediction, classification, optimization, and causal inference models — revenue forecasting, campaign performance prediction, user segmentation, spend allocation, personalization, and other ML projects across the team's portfolio Design and maintain models in both batch and online serving environments — owning the full path from research to production Evaluate, calibrate, and improve AI agent outputs — designing benchmarks, measuring accuracy, and identifying failure modes in AI-generated results Research and adopt new models and AI technologies to expand what the team can deliver — finding practical ways to scale our impact beyond what we can build manually Work with rich, large-scale user data — but the challenges here are deep: building models that stay accurate as the product and market shift, that scale reliably, and that keep driving decisions over time, not just at launch Extract signal from unstructured user data — prompts, photos, ad creatives, video interactions — using LLMs, embeddings, or custom approaches where domain-specific signal extraction matters Lead projects end-to-end: from identifying a business problem with stakeholders, through research and planning, to production deployment and monitoring Work directly with Finance, Marketing, and Prod
