jevframe: open-source library adds natural-language classification to pandas and Polars

_ScottCondron · x · 2026-09-22

Developer ktaletsk released jevframe, an open-source Python library built on TypeSafe's Jev model (early access via marimoio), bringing semantic AI to pandas and Polars DataFrames. It classifies text, analyzes sentiment, and scores rows via natural-language questions with full probability distributions. Results come back as ordinary Series/DataFrames preserving row order and indexes, with multi-question evaluation per row, bounded async concurrency, and optional caching. Use cases include labeling reviews, categorizing support tickets, flagging urgent messages, and rubric grading. GitHub repo includes an interactive marimo demo.

Original post →

More from coding & agent

coding & agent channel →