Infino

SQL, full-text, and vector search over your data on object storage — one engine, no server to run.

Infino keeps your data in Apache Parquet on object storage (local disk, Amazon S3, or any S3-compatible store) and runs SQL, full-text (BM25), and vector search over it from a single system. Each file is a valid Parquet file with BM25 and vector indexes embedded directly inside it; a table composes many such files with snapshot-isolated reads, append-only writes, and atomic commits. It runs in your process — there is no daemon, no cluster, and no managed service to operate.

Use it for RAG, agent memory, hybrid search, and semantic search: an embedded vector database, full-text (BM25) search engine, and SQL query engine in one library.

Installation

pip install infino

Or with uv:

uv add infino            # add to a uv-managed project
uv pip install infino    # install into the active environment

Requires Python 3.9 or newer. pyarrow is installed as a dependency; pandas is optional and used only if you pass DataFrames.

Quickstart

import infino
import pyarrow as pa

# Connect to a catalog. Use a local path or an S3 URI for durable storage;
# "memory://" is ephemeral and handy for tests.
db = infino.connect("./data")

# Tiny stand-in for your embedding model so this runs as-is — a 16-dim
# one-hot by topic. Real embeddings are dense and higher-dimensional.
def embed(topic):
    v = [0.0] * 16
    v[topic] = 1.0
    return v

# Declare a schema and which columns to index. An "_id" column is added
# automatically — you don't define it. A vector column's dim must be in
# [16, 4096]; here we use 16, the floor.
schema = pa.schema([
    pa.field("source", pa.large_utf8(), nullable=False),
    pa.field("body", pa.large_utf8(), nullable=False),
    pa.field("embedding", pa.list_(pa.float32(), 16), nullable=False),
])
docs = db.create_table(
    "docs", schema, infino.IndexSpec().fts("body").vector("embedding", 16, "cosine")
)

# Append rows. One append is one atomic commit.
docs.append([
    {"source": "help-center", "body": "To cancel a subscription, open Settings then Billing.", "embedding": embed(0)},
    {"source": "help-center", "body": "Refunds return to the original payment method.",         "embedding": embed(0)},
    {"source": "blog",        "body": "Enable dark mode under Settings then Appearance.",        "embedding": embed(1)},
])

# Retrieve context to ground an agent's next answer — keyword, vector,
# hybrid (BM25 + vector fused in one pass), or SQL. Each returns a pyarrow.Table.
keyword  = docs.bm25_search("body", "cancel subscription", k=5)                           # BM25
semantic = docs.vector_search("embedding", embed(0), k=5)                                 # vector kNN
hybrid   = docs.hybrid_search("body", "cancel subscription", "embedding", embed(0), k=5)  # fused
billing  = db.query_sql("SELECT body FROM docs WHERE source = 'help-center'")             # SQL filter

Documentation

Full docs, guides, and the API reference live at infino.ai/docs:

• Quickstart — install to first query
• Core concepts — superfiles, commits, and indexes
• Guides — Tables & indexing · Search: BM25, vector, hybrid · Embeddings · Storage & credentials
• SQL reference — query tables and the search table-valued functions
• API reference — the full Python surface, generated from the package
• MCP server — expose Infino retrieval to an AI agent (Claude, Cursor, …)

Building from source

The bindings are built with maturin and require a Rust toolchain.

python3 -m venv .venv && source .venv/bin/activate
pip install maturin pytest pyarrow
maturin develop          # compile the extension and install it into the venv
pytest tests/

License

Apache-2.0.