Skip to content
Open Source

The Zero-Dependency Hybrid Search Engine for Node.js

Blazing fast BM25 keyword search, smart regex filtering, and AI-powered hybrid — all in one lightweight package. From 100 records to 1M+, no external services required.

$ npm install @kevinsearch/core
< 10msSearch Latency
0Dependencies
1M+Records Supported
100%TypeScript

Built for Every Search Challenge

Three powerful search streams, plus extensible plugins — pick what you need.

Keyword Search

BM25 ranking with N-Gram typo tolerance. Index-based, fully offline, blazing fast at < 10ms.

Scan Search

Regex, Range, and Boolean filtering with absolute precision. Scans live data — no pre-indexing needed.

AI Hybrid Search

Vector Search (Cosine Similarity) merged with Keyword via RRF algorithm. Understands intent, not just keywords.

Synonym Expander

Zero-latency synonym expansion. Map domain-specific terms to boost recall without sacrificing speed.

Engagement Booster

Re-rank results by popularity, recency, or custom engagement signals to surface what matters most.

Infinite Scale

Swap in Postgres (tsvector + pgvector) adapter to handle 1M+ records — zero application code changes.

From Zero to Search in 3 Lines

No boilerplate. No configuration files. Just import, index, and search.

index.ts
1import { KeywordEngine } from '@kevinsearch/keyword'
2
3const engine = new KeywordEngine()
4await engine.index(documents)
5
6const results = await engine.search({
7 raw: 'senior react developer',
8 pagination: { page: 1, pageSize: 10 }
9})
10
11// results.items — ranked by BM25 relevance

Interactive Playground

Experience @kevinsearch in action — try different search modes across multiple data domains.

Search developers, products, videos...
Playground Coming Soon

We're building an interactive demo where you can test keyword, scan, and AI hybrid search across datasets like developer profiles, e-commerce products, and more.

Ready to Supercharge Your Search?

Get started in minutes. Free and open source.