Anthony Wu Email & Phone Number

California, United States

San Francisco, California, United States

1. Recommender Systems- Built a batch mode recommender system from scratch- Data input pipeline with customer website activity data, customer attributes, and product attributes using Snowflake/Polars/Pandas/Numpy- Trained multiple Matrix Factorization models with hyperparameter optimization using Ray Tune and Aim, and model versioning with Mlflow- Developed an inferencing pipeline for email product recommendations, leading to roughly 13-15% improvement in revenue per customer.- Used FAISS to generate similar product recommendations for email campaigns, resulting in 5 - 10% improvement in revenue per customer.2. Uplift Modeling- Built a causual inference model using Uber's CausalML package to improve promotion assignment strategy by roughly 20% 3. Customer Intelligence- Customer Active Shopper Journeys: Developed a data processing pipeline to separate website events into customer journeys to optimize email timing and delivery. Collaborated with several stakeholders on the email strategy and data engineering teams to build around a new email framework. Resulted in a 10 - 20% improvement in conversion rate and revenue per email user.4. Analytics- eComm Forecasting: Worked with Marketing and US eCommerce teams to generate weekly and monthly forecasts using pmdarima for interpretability, lowering error rate by X%.- Product Clustering: Deployed a product clustering model based on sales and trend and collaborated with the Google Shopping ads team to improve conversion rates on popular/trending products by around 75%. 5. MLOps- Built out a new resource manager in Dagster to use .parquet files instead of pickle files, decreasing AWS storage usage by 30 - 50% depending on the file contents- Managed the transition of the entire DS / ML codebase from AWS managed Airflow to Dagster.- Fully deployed/productionalized around 20 separate models/analytical solutions using Dagster / Airflow.