This study uses polars [0][1] -- a new Data Science library written in Rust (core) with bindings to Python and Javascript. Perfect replacement for pandas. Polars' claims to fame are speed (faster than pretty much anything out there [2][3]), low memory pressure (very visible when comparing to pandas) and consistent API across Rust, Python, JS.
[0] https://www.pola.rs/
[1] https://github.com/pola-rs/polars
[2] https://h2oai.github.io/db-benchmark/
[3] https://www.pola.rs/benchmarks.html