Associate Machine Learning Engineer
Los Angeles, California, Us
• Conceptualized, developed, and deployed company's first ML-driven ranking/recommender system (XGBoost model) within 9 months, leveraging AWS for preprocessing and inference, H2O.ai for training, and CometML for monitoring and experimentation• Drove a 50%+ increase in mobile app shift impressions within the first month of recommender's launch• Led end-to-end development of: - Airflow & Glue pipelines, for building viable classification training datasets (TB sized) to RDS - SageMaker & CometML training pipelines, for fast experimentation, reporting, and tuning - Training, tuning, testing, and releasing a recommender model from 0 data to prod - Lambda & SQS inference pipelines, to produce (fan-out) millions of near-real time predictions (candidate generation, feature engineering, invoke model, relay to backend)