Worked on the pola-rs/polars repository to enhance the stability of time-based DataFrame operations, specifically addressing issues with empty inputs. Focused on refining the dt.truncate and dt.round methods to ensure they handle empty DataFrames without crashing and consistently return the correct schema and shape. Utilized both Python and Rust to implement robust input validation and comprehensive regression tests, verifying correct behavior across edge cases. This targeted bug fix improved reliability for production data pipelines that depend on Polars’ time-based features. The work demonstrated careful attention to data manipulation, testing, and cross-language integration within a single, production-focused codebase.