Ml4Db Research Assistant
- Developed ML system for query cardinality estimation that works across DBMSs using few-shot transfer learning- Refined Python preprocessing pipeline to extract, create, and standardize features- Identified a stacked model architecture that improves performance on certain classes of query nodes by ~20%- Wrote bash scripts to handle generating TPC-H data for model training- Surveyed literature to understand existing methods to represent query plans and operators for ML models- Onboarded 2 new team members by creating documentation and explaining codebase