6+ years in product management, with 3+ years owning a recommendation, ranking, personalization, or relevance system in production.
Deep understanding of experiment design and inference, including minimum detectable effect, statistical power, sample ratio mismatch, A/A testing, multiple-comparison correction, and pre-registration. You can tell when a readout will not support the claim being made from it, and you say so.
Working knowledge of the modelling itself: classification and regression, probability calibration, predicted conversion rate, expected-value objectives, and how the choice of objective changes model behavior.
Hands-on experience with ML and data platforms such as Snowflake, BigQuery, Spark, Airflow, dbt, MLFlow, Vertex AI, and feature stores.
Proven record of moving model changes into production traffic through a controlled experiment process, rather than shipping on a dashboard reading.
Knowledge of performance advertising economics and ad tech ecosystems, including eCPM, RPU, bid multipliers, campaign hierarchies, and attribution models.
SQL you write yourself, and enough comfort reading application code to verify a claim instead of taking it on trust.
Excellent written communication and stakeholder management across technical and non-technical audiences. Specs, decision records, and readouts are the output of this role, and other teams act on them without you in the room.
Bachelor’s or Master’s in Computer Science, Engineering, Statistics, Economics, or a related field.