Rushikesh Nimkar's Portfolio

A modern portfolio website built with Next.js 15, featuring AI chat integration and interactive animations.

✨ Features

• AI-Powered Email Generation: Create professional emails with AI assistance
• Interactive Chat: Talk directly with an AI version of me using advanced LLMs
• Vector Embeddings: Semantic search capabilities for more accurate AI responses
• Dynamic Animations: Engaging UI with smooth animations
• Responsive Design: Seamless experience across all devices
• Dark Mode: Eye-friendly interface for all lighting conditions
• Project Showcase: Interactive displays of my work and contributions
• AI-Powered Dynamic Theming: Transform the entire interface with natural language requests, allowing real-time visual customization

---

🛠 Tech Stack

Layer	Technologies
Frontend	Next.js 15, TypeScript, Tailwind CSS, Framer Motion
Backend	Node.js, Next.js API Routes
AI Integration	OpenRouter API
Embeddings	Google Gemini
Vector Database	Pinecone
Search	Tavily API
Email	Nodemailer, Abstract API
Deployment	Vercel + Edge Functions

---

Model Status

 

This badge shows the real-time operational status of selected AI models from OpenRouter.

Each badge displays:

• Model name: The model's identifier
• Status:
  • operational – model is healthy (uptime ≥ 90%)
  • Degraded – degraded performance (uptime ≥ 50%)
  • offline – not available (uptime < 50%)
• Uptime: The average uptime over the past 30 minutes

Clicking the badge opens the live SVG endpoint.

⚠️ Note:
A Degraded does not mean the model is completely unusable.
It simply indicates that some requests may fail intermittently due to degraded uptime or performance.

This badge shows the real-time operational status of the Kimi model.

---

🚀 Getting Started

⚠️ IMPORTANT: Before starting, make sure to set up all required environment variables in a .env file as described in the Environment Variables section below. The application will not work properly without these configurations.

1. Clone and install:

git clone https://github.com/Rushikeshnimkar/portfolio2025.git
cd portfolio
npm install

2. Update your personal information:

# Edit the components/character/character.ts file with your own details
# This is crucial for the AI to represent you correctly

3. Run development server:

npm run dev

Note: The first run will take longer than usual as the system creates and saves your embeddings to the vector database. Take this time to relax and enjoy a coffee! ☕

4. Open http://localhost:3000

⚠️ IMPORTANT NOTE: Make sure to uncomment the following line in api/chat/route.ts & api/theme/route.ts when working in development:

// "http://localhost:3000",

📧 Email Generation Feature

The portfolio includes an AI-powered email generation system that:

• Creates professional emails based on simple prompts
• Validates sender email addresses
• Supports both AI-generated and manual email composition
• Features elegant text generation animations

💬 AI Chat Integration

Chat with an AI version of me that:

• Answers questions about my skills, experience, and projects
• Accesses real-time information when needed via Tavily search
• Maintains conversation context across messages
• Provides accurate information about my background and expertise
• Uses vector embeddings for more accurate and relevant responses

🎨 AI Theme Customization

The portfolio includes an AI-powered theme customization system that:

• Transforms the UI with natural language commands
• Understands specific page elements through vector embeddings
• Applies changes in real-time with visual feedback
• Generates custom JavaScript on-the-fly for unique visual effects
• Maintains responsive design across all device sizes
• Supports both page-specific and site-wide modifications

🔍 Vector Database Integration

The portfolio uses Pinecone vector database to store and retrieve embeddings:

• Semantic search capabilities for more accurate AI responses
• Google Gemini embeddings for high-quality vector representations
• Efficient retrieval of relevant information based on user queries
• One-time initialization with profile data for persistent storage

🔒 Security Features

• CORS protection for API routes
• Email validation to prevent spam
• Rate limiting on sensitive endpoints

Environment Variables

Create a .env file in the root directory and add your environment variables:

# Development environment (all the apis are free)
EMAIL_USER="your emailid"
EMAIL_APP_PASSWORD="your App Password" # get it from your Google Account settings
ABSTRACT_API_KEY="your abstract api key" # https://app.abstractapi.com/
OPENROUTER_API_KEY="your openrouter api key" # https://openrouter.ai/
TAVILY_API_KEY="your Tavily api key" # https://tavily.com/

# Vector Database (Pinecone)
PINECONE_API_KEY="your Pinecone API key" # https://app.pinecone.io/
PINECONE_INDEX_NAME="your index name" # e.g., portfolio-embeddings

# Google Gemini (for embeddings)
GOOGLE_API_KEY="your Google API key" # https://ai.google.dev/

#jwt auth
JWT_EXPIRY="1m"
JWT_SECRET="your jwt token" # use this command for linux = "openssl rand -hex 64" or node -e "console.log(require('crypto').randomBytes(32).toString('hex'))"  (works on windows/linux/mac)

Vector Store Configuration

The vector store is configured in lib/embeddings.ts with the following settings:

• Dimension: 768 (Google Gemini embeddings)
• Metric: Cosine similarity
• Pod type: Starter (free tier)

You can modify these settings in the initializeVectorStore function if needed.

📸 Preview

 

---

📄 License

This project is open source under the MIT License.

---

🙋‍♂️ Author

Rushikesh Nimkar
🔗 rushikeshnimkar.xyz
📧 rushikeshnimkar396@gmail.com