During this period, contributed to ParadeDB by implementing BM25 index pushdown support for the PostgreSQL ltree <@ operator, enabling hierarchical read queries to utilize the BM25 index directly and bypass heap-side filtering. This work involved adding checks and conversions to ensure correct indexing behavior and query execution, particularly for queries such as WHERE category <@ 'Top.Science'::ltree. The feature, delivered in the paradedb/paradedb repository, addressed performance bottlenecks for hierarchical data access. The development process relied on Rust and SQL, with a focus on database optimization and query performance tuning, and included collaborative code review across teams to ensure robustness.