Skip to main content

Apiary

A distributed data processing framework inspired by bee colony intelligence

  • Runs on Small Hardware

    Designed for Raspberry Pis, NUCs, and old laptops. Not a cloud system squeezed onto small hardware, but a system built for resource-constrained devices that scales to the cloud.

  • Zero-Configuration Multi-Node

    Add a node by pointing it at the same storage bucket. No seed nodes, no tokens, no mesh network setup. Nodes discover each other through the storage layer.

  • Biology-Driven Design

    Memory budgets, cell sizing, backpressure, and failure recovery are governed by bee-inspired behavioral algorithms. Mason bee isolation, leafcutter sizing, and colony temperature are not metaphors but runtime mechanisms.

  • SQL + Python

    Query with Apache DataFusion SQL, write with Python. Apache Arrow provides zero-copy interop between the Rust engine and the Python SDK.

  • Object Storage as Truth

    All data lives in S3, MinIO, GCS, or local filesystem. Nodes are stateless workers. Lose a node and lose nothing. The bucket is the complete system of record.

  • ACID Transactions

    Ledger-based transactions using conditional writes, inspired by Delta Lake. No Raft, no consensus protocol. The storage layer provides the atomic compare-and-swap.