• Building ML infra.

• Built data infra, e.g., data warehouse, unified stream and batch processing system.

• Built ML infra, e.g., candidate retrieval, feature store, model serving.

• Built end-to-end ML systems for fraud detection and product recommendation.

• Built centralized product catalogs to make Bolt the source of truth for all products from all merchants.

• Backend, scalability, data engineering, and machine learning. Tools include Java, Spark, and Python.

• Built end-to-end ML system for CTR prediction.

• Built ad retrieval and indexing systems.

• Built ML infra.

• Built data pipelines and tools.

• Led analytical efforts that provide actionable insights to support business operations and decision-making.

• Developed algorithms for data de-identification that effectively prevent individuals from being identified in order to protect privacy, while allowing data users to perform inferences accurately as if they had access to the original data.

• Conducted research on computer vision with focus on object classification and detection for images. Developed deformable template models for objects and the corresponding estimation algorithms. Achieved competitive classification rates using small training sets.