Machine Learning Engineer

CARTA's group develops its own advertising products, including a demand-side platform (DSP) that bids in real-time ad auctions, with a current focus on performance ads aimed at app installs and purchases. The Machine Learning Engineer works on the bid-optimisation logic in a full-cycle data science team, which runs from data analysis and hypotheses through A/B tests in production.

Tasks include exploratory analysis of bid logs, redesigning logs and using external data, building models such as click and conversion prediction, building bid-price optimisation with mathematical optimisation and control engineering, hypothesis testing including online experiments, building and running the analysis platform and ML pipelines for large data, and tracking academic research. The data is large and imbalanced, and user response is visible only for ads that won the auction. The stack includes Python, AWS, dbt, Snowflake, Prefect and Terraform.

Applicants need five or more years implementing and operating machine learning models in Python, end-to-end experience from hypothesis building to A/B test evaluation, and large-scale aggregation with Snowflake or BigQuery. Advertising domain knowledge, recommender or ranking experience, causal inference and competition results are preferred.