Polars

If you're still using pandas out of habit, this is your nudge to try Polars. It's built on Apache Arrow and genuinely faster, especially once your datasets get big enough to hurt. The expression API is cleaner than pandas (no more lambda chaining), lazy evaluation means you can scan giant CSVs without loading them entirely into memory, and operations parallelize by default. The skill covers the core patterns: select, filter, group_by, window functions with over(), and practical stuff like when to use LazyFrame versus DataFrame. The pandas migration reference is honest about the differences, no index, strict typing, different syntax, which saves you from discovering them the hard way at 2am.