Ruby bindings to the Rust Polars DataFrame library. In the wroclove.rb 2026 CSV-parsing benchmark it gave the biggest speedup — roughly 22× faster than the Ruby stdlib CSV on 1M rows, and ~100MB post-load memory on a 30MB CSV (vs 400MB for stdlib Ruby CSV). Supports a low-memory flag that trades some speed for a smaller memory footprint. Used in production by the speaker to parse ~300MB CSVs within a single web request without caching or pre-calculation. Relatively unknown in the Ruby community at the time of the talk (about two hands raised in a conference-room audience).