Skip to content

PyChassis

PyChassis provides high-performance Python bindings for the Chassis vector storage engine.

It combines the safety and raw speed of Rust with the ease of use of Python. Unlike many vector libraries that run primarily in-memory, Chassis is designed for embedded, on-disk persistence first.

Key Features

  • Zero-Copy Search: Vectors are memory-mapped, allowing instant access to datasets larger than RAM.
  • Crash Safety: ACID-compliant persistence guarantees your data is safe even if the process is killed.
  • Standard Interface: Fully compatible with NumPy arrays.
  • No Server Required: Runs entirely in-process. No Docker containers or external services to manage.

Installation

PyChassis requires Python 3.8+ and a compatible OS (Linux, macOS, Windows).

pip install pychassis

import chassis
import numpy as np

# 1. Open an index (creates file if not exists)
index = chassis.VectorIndex("my_vectors.chassis", dimensions=128)

# 2. Add some data
vector = np.random.rand(128)
vec_id = index.add(vector)

# 3. Persist to disk
index.flush()

# 4. Search
results = index.search(vector, k=5)
print(f"Found ID: {results[0].id} with distance: {results[0].distance}")