Standalone S3 & Vector Engine

Object Storage & Vector Search
In One 15MB Binary

Pranor Vault provides full AWS S3 API compatibility alongside native HNSW vector similarity search. High throughput, zero external dependencies, zero maintenance.

Quick Start View on GitHub

Standalone Install: docker run -p 8084:8084 vyuvaraj/pranor-vault:latest

Core Capabilities

Built for High-Performance Cloud & AI Applications

Why manage separate S3-compatible clusters and vector databases when you can run both in a single lightweight Go binary?

🪣

S3 API Compatibility

Compatible with standard AWS S3 SDKs, MinIO tools, and CLI commands. Supports Multipart uploads, Bucket Policies, and Presigned URLs.

🧠

Native HNSW Vector Index

Store, index, and query high-dimensional embeddings using Cosine, Euclidean, or Dot-product distances with millisecond latency.

⚡

Zero-Allocation Streaming

Optimized buffer reuse engine delivers 10Gbps+ read throughput with minimal memory footprint under concurrent access.

Comparison Matrix

Vault vs MinIO & Qdrant

See how Pranor Vault simplifies your infrastructure stack without sacrificing features.

Feature	MinIO + Vector DB	Pranor Vault
Single Binary Deployment	✕ (Requires multiple services)	✓ (Single 15MB binary)
AWS S3 API Support	✓	✓
Native Vector Search (HNSW)	✕ (Requires Qdrant/Pinecone)	✓ Built-in
Memory Consumption	High (500MB+ per node)	Ultra-Low (~25MB baseline)
Direct Vector-Object Correlation	✕ (Separate metadata DB)	✓ Direct linkage
Open Source (AGPL-3.0)	✓	✓

Code Examples

Use from Any Language or Framework

Pranor Vault works natively with Python, TypeScript/Node.js, Go, cURL, or directly in Pranor applications.

import boto3
import requests

# 1. Standard AWS S3 SDK for object upload
s3 = boto3.client('s3', endpoint_url='http://localhost:8084')
s3.upload_file('report.pdf', 'documents', 'report.pdf')

# 2. HNSW Vector Similarity Query via REST
res = requests.post('http://localhost:8084/api/v1/vectors/search', json={
    "collection": "doc-embeddings",
    "vector": [0.12, -0.44, 0.89, 0.05],
    "top_k": 5
})
print("Top match:", res.json()['results'][0]['key'])
  

import { S3Client, PutObjectCommand } from "@aws-sdk/client-s3";

// 1. Standard S3 Client
const s3 = new S3Client({ endpoint: "http://localhost:8084", region: "us-east-1" });
await s3.send(new PutObjectCommand({ Bucket: "documents", Key: "report.pdf", Body: fileBuffer }));

// 2. Vector Similarity Search
const res = await fetch("http://localhost:8084/api/v1/vectors/search", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ collection: "doc-embeddings", vector: [0.12, -0.44, 0.89], top_k: 5 })
});
const data = await res.json();
console.log("Matches:", data.results);
  

package main

import (
    "context"
    "github.com/aws/aws-sdk-go-v2/aws"
    "github.com/aws/aws-sdk-go-v2/service/s3"
)

func main() {
    // Standard AWS Go SDK v2 initialized against Pranor Vault
    client := s3.NewFromConfig(cfg, func(o *s3.Options) {
        o.BaseEndpoint = aws.String("http://localhost:8084")
    })
    client.PutObject(context.TODO(), &s3.PutObjectInput{
        Bucket: aws.String("documents"),
        Key:    aws.String("report.pdf"),
    })
}
  

# Upload object via standard S3 API
curl -X PUT http://localhost:8084/documents/report.pdf --data-binary @report.pdf

# Execute HNSW Vector Search
curl -X POST http://localhost:8084/api/v1/vectors/search \
  -H "Content-Type: application/json" \
  -d '{"collection": "doc-embeddings", "vector": [0.12, -0.44, 0.89], "top_k": 5}'
  

import "std/vault"

fn main() {
    // Native Pranor syntax
    vault.put_object("documents", "report.pdf", data)
    let results = vault.search_vector("doc-embeddings", query_embedding, 5)
    log.info("Top match: " + results[0].key)
}
  

Object Storage & Vector Search In One 15MB Binary