🔍 AI Semantic Search

Powered by Endee Vector Database

A production-quality semantic search engine that understands meaning, context, and intent — not just keywords.

🚀 Try Search · 📖 API Docs · ⭐ Star Endee · 🍴 Forked Repo

---

📌 Overview

AI Semantic Search is a full-stack web application that demonstrates real-world semantic search using Endee as the vector database backend.

Unlike traditional keyword search, this system converts natural language queries into 128-dimensional vectors and finds conceptually similar documents using cosine similarity — even when the documents share no common vocabulary with the query.

Built as part of the Tap Academy × Endee.io evaluation assignment — March 2026.

---

✨ Features

Feature	Description
🧠 Semantic Search	Understands meaning and intent, not just keywords
⚡ Endee Vector DB	Real vector similarity search via Docker + npm SDK
🎯 Ranked Results	Top 3 results with cosine similarity confidence scores
📊 Category Filter	Filter results by technology, science, philosophy, etc.
🕐 Search History	Remembers your recent searches locally
📋 Copy & Share	Copy results or share search links instantly
📡 Animated UI	Radar loader, typing effect, floating orbs, hover animations
📖 API Docs Page	Full documentation at /docs
📈 Stats Dashboard	Analytics overview at /stats
🌐 About Page	Project overview, timeline and tech stack at /about

---

🏗️ System Architecture

User Query (Natural Language)
          │
          ▼
┌─────────────────────────┐
│   Next.js Frontend      │  ← React 18 + Tailwind CSS
│   Space Grotesk UI      │     Animations, Typing Effect
└──────────┬──────────────┘
           │
           ▼
┌─────────────────────────┐
│   /api/search Route     │  ← Next.js App Router
│   (Backend Logic)       │     Edge-ready API
└──────────┬──────────────┘
           │
           ▼
┌─────────────────────────┐
│   Text → Vector         │  ← Character frequency
│   Embedding             │     128-dimensional space
└──────────┬──────────────┘
           │
           ▼
┌─────────────────────────┐
│   Endee Vector DB       │  ← Docker container
│   localhost:8080        │     Cosine similarity
└──────────┬──────────────┘
           │
           ▼
   Ranked Results
   with Scores & Metadata

---

🧠 How Endee is Used

Property	Value
Index name	semantic_search
Dimensions	128
Space type	cosine
Documents	15 knowledge nodes
SDK	Official endee npm package
Query	index.query({ vector, topK: 3 })
Host	http://localhost:8080

Integration Code

import { Endee } from 'endee'

// 1. Connect to Endee
const client = new Endee()
client.setBaseUrl('http://localhost:8080/api/v1')

// 2. Create index
await client.createIndex({
  name: 'semantic_search',
  dimension: 128,
  spaceType: 'cosine',
})

// 3. Upsert documents as vectors
const index = await client.getIndex('semantic_search')
await index.upsert(documents.map(doc => ({
  id: doc.id,
  vector: textToVector(doc.title + ' ' + doc.snippet),
  meta: { title, snippet, tags, category }
})))

// 4. Query with vector similarity
const results = await index.query({
  vector: textToVector(userQuery),
  topK: 3
})
// results[0].similarity = cosine similarity score (0-1)
// results[0].meta = document metadata

---

⚙️ Tech Stack

Layer	Technology	Purpose
Frontend	Next.js 14, React 18	UI & routing
Styling	Tailwind CSS	Utility styling
Typography	Space Grotesk, Plus Jakarta Sans, Space Mono	Premium fonts
Backend	Next.js API Routes	Server logic
Vector DB	Endee (Docker)	Semantic search
Embeddings	Character frequency vectors (128-dim)	Text → vector
SDK	Official endee npm package	Endee integration

---

🛠️ Setup Instructions

Prerequisites

• Node.js 18+
• Docker Desktop

1. Clone the repository

git clone https://github.com/rakeshits/endee
cd endee

2. Start Endee Vector Database

docker run -d -p 8080:8080 \
  -e NDD_AUTH_TOKEN="" \
  --name endee-server \
  endeeio/endee-server:latest

Verify it's running: docker ps — you should see endee-server

3. Install dependencies

npm install

4. Run the development server

npm run dev

5. Open in browser

http://localhost:3000

On first search, the app automatically creates the Endee index and upserts all 15 documents as vectors.

---

📁 Project Structure

📦 endee/
├── 📂 app/
│   ├── 📄 page.js                 # Landing page with hero
│   ├── 📂 search/
│   │   └── 📄 page.js             # Search interface
│   ├── 📂 about/
│   │   └── 📄 page.js             # About & timeline
│   ├── 📂 stats/
│   │   └── 📄 page.js             # Analytics dashboard
│   ├── 📂 docs/
│   │   └── 📄 page.js             # API documentation
│   └── 📂 api/search/
│       └── 📄 route.js            # Endee vector search API
├── 📂 components/
│   ├── 📄 Navbar.js               # Sticky navigation
│   ├── 📄 Footer.js               # Site footer
│   ├── 📄 SearchBar.js            # Search input component
│   └── 📄 ResultCard.js           # Result card with actions
├── 📄 app/globals.css             # Design system & animations
├── 📄 tailwind.config.js
├── 📄 package.json
└── 📄 README.md

---

🔍 API Reference

Endpoint

GET /api/search?query=<string>

Example Request

curl "http://localhost:3000/api/search?query=neural+networks"

Example Response

{
  "query": "neural networks",
  "results": [
    {
      "id": "doc1",
      "title": "Neural Networks & Deep Learning",
      "snippet": "Artificial neural networks are computing systems...",
      "tags": ["AI", "machine learning", "neural"],
      "category": "technology",
      "score": 0.923,
      "source": "endee"
    }
  ],
  "totalFound": 3,
  "searchTime": "28ms",
  "engine": "Endee Vector Database",
  "endee": {
    "index": "semantic_search",
    "dimensions": 128,
    "spaceType": "cosine",
    "host": "localhost:8080"
  }
}

Response Fields

Field	Type	Description
query	string	Original search query
results	array	Top 3 ranked results
score	number	Cosine similarity (0–1)
source	string	endee or fallback
engine	string	Search engine used
searchTime	string	Query processing time

---

📊 Knowledge Base

15 curated documents across 7 categories:

Category	Documents
🔵 Technology	Neural Networks, Quantum Computing, LLMs, Blockchain, Semantic Search
🟢 Science	CRISPR, Evolutionary Biology, Renewable Energy
🟣 Space	Black Holes, Mars Missions
🩵 Health	Mindfulness, Human Microbiome
🟡 Philosophy	Consciousness
🟤 Nature	Climate Change
🔴 History	Ancient Rome

---

🗺️ Pages

Page	Route	Description
🏠 Home	/	Hero, stats, features, how it works
🔍 Search	/search	Main search interface
📖 About	/about	Project overview & timeline
📈 Stats	/stats	Analytics dashboard
📄 Docs	/docs	Full API documentation

---

🎨 Design System

Token	Value	Usage
--p	#7c6dff	Primary purple accent
--pk	#ff6b9d	Pink accent
--g	#00d4aa	Green / success
--bg	#050508	Page background
--tx	#f0eeff	Primary text
Display font	Space Grotesk	Headings
Body font	Plus Jakarta Sans	UI text
Mono font	Space Mono	Labels, code

---

🔗 Links

Resource	URL
⭐ Star Endee	github.com/endee-io/endee
🍴 Forked Repo	github.com/rakeshits/endee
📖 Endee Docs	docs.endee.io
🌐 Endee Website	endee.io

---

🏗️ Build Timeline

Day 1 AM  →  Forked Endee repo, setup Next.js + Tailwind CSS
Day 1 PM  →  Built semantic search engine with 15 documents
Day 2 AM  →  Integrated Endee vector DB via Docker + npm SDK
Day 2 PM  →  Built full UI — hero, search, about, stats, docs
Day 2 PM  →  Added animations, typing effect, search history
Day 2 EOD →  Pushed to GitHub, submitted to Tap Academy

---

Built with ❤️ using Endee Vector Database

Tap Academy × Endee.io Evaluation Assignment — March 2026