Looking for contributors — SlothDB is beating DuckDB, let's make it a project people reach for

[SouravRoy-ETL]

Short version: SlothDB now beats DuckDB 1.1–8.6× on every query in the benchmark suite across 7 file formats (Parquet, CSV, JSON, Avro, Arrow, SQLite, Excel). The engine works. What I'd like help with is turning it into a project people actually reach for.

I'm a solo maintainer. Pushing this forward on evenings and weekends has a ceiling, and I've hit it. Writing this to open the door.

What's there today

• Vectorized columnar engine, morsel-driven parallelism, fused scan+aggregate
• 7 file format readers in core, no extensions required
• Python package on PyPI (pip install slothdb), single-binary CLI, stable C ABI for C/C++ embedding
• 326 tests, 131k+ assertions, green CI on Windows, Linux, macOS
• MIT licensed

Where I'd love help

Rough priority. Pick whatever looks interesting — not a checklist, just the shape of what's open.

• ClickBench submission. Port the 43-query ClickBench suite to SlothDB, add a benchmark.sh, open a PR against ClickHouse/ClickBench. This is the single highest-leverage credibility unlock for the project.
• SQL dialect gaps. There are corners that still surprise users — certain window function edge cases, correlated subqueries in weird positions, a handful of date/time functions missing. Good first-contributor territory; each one is usually a self-contained patch.
• R and Node.js bindings. Python exists and works well. The C ABI is stable enough that wrapping it in another language is mostly mechanical.
• HTTP and S3 file readers. Currently everything is local filesystem. SELECT * FROM 'https://...' and 's3://bucket/...' are on the roadmap but not started.
• Excel write path. Reads work. COPY TO 'file.xlsx' doesn't yet.
• Docs. A proper function-by-function SQL reference page. Examples gallery. Tutorial-style "day in the life of" posts. The current docs are terse.
• Bug reports from real workloads. If you try SlothDB on an actual pipeline and something breaks, that's the most useful contribution there is — a failing test case is worth a weekend of my time.

How to start

• Repo and build instructions: https://github.com/SouravRoy-ETL/slothdb
• Full docs: https://github.com/SouravRoy-ETL/slothdb/blob/main/docs/DOCUMENTATION.md
• I'll seed good-first-issue labels in the next few days

Drop a comment below if any of the above looks interesting, or just open an issue or draft PR directly. I respond fast — usually same day.

It's v0.1.3, about six months old, one person. The engine is real. What comes next is up to whoever shows up.