GitHub: https://github.com/datumbrain/otters

I started this because most Go DataFrame libraries I tried were either abandoned, hard to use, or didn’t feel idiomatic to Go. I wanted something that:

• works fully in-process (no cluster / Spark dependency)

• uses native Go types instead of generic interface blobs

• supports expressive chained operations

• stays readable and explicit like Go code should

• can scale to large datasets in memory pipelines

Design goals were:

• simplicity over magic

• type safety over dynamic guessing

• composability for real data pipelines

It’s still early but functional, and I’d genuinely appreciate feedback from people who’ve built data tooling or worked on columnar engines / query planners. Critiques welcome — API design, performance, architecture, anything.