Pure Node.js · Zero Dependencies · Columnar Engine

DATA
PROCESSING
REIMAGINED

Columnar data engine for Node.js. SharedArrayBuffer + Worker Threads. Zero dependencies. Faster than Polars on load.

$ npm i octopus-data
GitHub
—
Weekly Downloads
3.44s
Load 7.8M rows
0
Dependencies
MIT
License
Performance

BENCHMARK
7,832,546 ROWS

EngineRelative timeTotal
octopus-data
5.02s
Polars (Python)
5.95s
Pandas (Python)
25.80s
CSV · 7,832,546 rows · Feature Engineering + Filtering
Octopus uses SharedArrayBuffer + Worker Threads.
Ecosystem

THE
OCTOPUS FAMILY

● Live
octopus-data

Columnar data engine. Load, filter, group, aggregate CSV/JSON with zero overhead.

CSVJSONGroupByWorker ThreadsSAB
$ npm i octopus-data
○ Coming Soon
octopus-decomposer

Binary format reader: Parquet, Arrow IPC, Avro. Pure Node.js, no native modules.

ParquetArrowAvro
$ npm i octopus-decomposer
Architecture

BUILT
DIFFERENT

01
ZERO DEPENDENCIES

Node.js core only. No npm cascade, no hidden CVEs.

02
COLUMNAR STORAGE

Float64Array + TypedArrays. Cache-efficient aggregations.

03
SHARED MEMORY

SharedArrayBuffer across Worker Threads — no copies.

04
PARALLEL BY DEFAULT

Automatic dataset partitioning, every CPU core used.