# Binal Digital Twin Knowledge Base
This directory contains sample data and scripts for populating the Upstash Vector database with Binal's professional information.
## Sample Data Structure
Each entry in the vector database should have:
- **text/content**: The main content to be searched
- **metadata**: Additional information like source, category, etc.
- **vector**: The embedding (generated automatically by Upstash)
## Example Data Points
```typescript
const sampleData = [
{
id: "education-1",
text: "Binal has a Master's degree in Computer Science with specialization in Machine Learning and Artificial Intelligence. Graduated with distinction from a top-tier university.",
metadata: {
category: "education",
source: "Academic Records",
importance: "high"
}
},
{
id: "skills-python-1",
text: "Binal is highly proficient in Python programming with 5+ years of experience. Expert in libraries like pandas, numpy, scikit-learn, TensorFlow, and PyTorch for data science and machine learning projects.",
metadata: {
category: "technical_skills",
source: "Professional Portfolio",
skills: ["Python", "Machine Learning", "Data Science"],
importance: "high"
}
},
{
id: "projects-rag-1",
text: "Binal built a sophisticated RAG (Retrieval-Augmented Generation) system using vector databases, embedding models, and large language models. The project demonstrates expertise in modern AI architectures.",
metadata: {
category: "projects",
source: "GitHub Portfolio",
technologies: ["RAG", "Vector Databases", "LLMs", "Embeddings"],
importance: "high"
}
},
{
id: "experience-ml-1",
text: "Binal has extensive experience in machine learning model development, from data preprocessing to model deployment. Worked on various ML projects including NLP, computer vision, and predictive analytics.",
metadata: {
category: "experience",
source: "Work Portfolio",
domain: ["NLP", "Computer Vision", "Predictive Analytics"],
importance: "high"
}
}
];
```
## Setup Instructions
1. Create an Upstash Vector database at https://upstash.com/
2. Get your REST URL and Token
3. Add them to your `.env.local` file:
```
UPSTASH_VECTOR_REST_URL=your_url_here
UPSTASH_VECTOR_REST_TOKEN=your_token_here
```
4. Use the Upstash console or API to upload your data
5. Test the connection using the MCP server interface
## Data Population Script
You can create a simple script to populate the database:
```typescript
import { Index } from "@upstash/vector";
const index = new Index({
url: process.env.UPSTASH_VECTOR_REST_URL!,
token: process.env.UPSTASH_VECTOR_REST_TOKEN!,
});
async function populateDatabase() {
for (const item of sampleData) {
await index.upsert({
id: item.id,
data: item.text,
metadata: item.metadata,
});
}
}
```
## Tips for Better Search Results
1. **Comprehensive Coverage**: Include information about education, skills, projects, experience, certifications, etc.
2. **Clear Descriptions**: Write clear, specific descriptions that answer common questions
3. **Rich Metadata**: Include categories, importance levels, and relevant tags
4. **Natural Language**: Write in natural language that matches how people ask questions
5. **Regular Updates**: Keep the knowledge base updated with new experiences and skills
