🌌 NEXUS — Multimodal AI Research Platform

NEXUS is an autonomous, multi-agent research system designed to search, analyze, synthesize, and critique information in real-time. By leveraging high-speed inference and a parallelized agentic workflow, NEXUS generates comprehensive research reports with full citations and objective critiques.

🚀 Key Features

• Real-Time SSE Streaming: Watch the agents think and write with a "ChatGPT-like" streaming experience.
• Autonomous Multi-Agent Pipeline:
  • Search Agent: Finds the most relevant sources via Tavily AI.
  • Scraper Agent: Extracts high-quality content from web sources.
  • Writer Agent: Synthesizes data into a professional markdown report.
  • Critic Agent: Evaluates the report for quality and objectivity.
• Premium UI/UX: Modern, glassmorphic design with interactive particle physics and smooth transitions.
• Multimodal Support: Upload documents or images for contextualized research.
• Hybrid Cloud Deployment: Optimized for speed (Vercel) and persistent streaming (Render).

---

🛠️ Technology Stack

• Frontend: Vanilla JS, HTML5, CSS3 (Glassmorphism, Canvas API)
• Backend: FastAPI (Python 3.9+)
• LLM Inference: Groq LPU (Llama-3-70b/8b)
• Search Engine: Tavily AI
• Vision: Google Gemini 1.5 Pro
• Hosting: Vercel (Frontend), Render (Backend)

---

💻 Local Setup

1. Clone the Repository

git clone https://github.com/SparshGarg999/nexus.git
cd nexus

2. Configure Environment Variables

Create a .env file in the root directory:

GROQ_API_KEY=your_groq_key
TAVILY_API_KEY=your_tavily_key
GEMINI_API_KEY=your_google_ai_key
PORT=8000

3. Install Dependencies

pip install -r requirements.txt

4. Run the Application

python run.py

Visit http://localhost:8000 to access the platform.

---

☁️ Deployment Guide

Backend (Render)

1. Push this repo to GitHub.
2. Go to Render -> New -> Blueprint.
3. Connect this repository. Render will use the render.yaml to automatically configure the service.
4. Provide your API keys when prompted.
5. Set FRONTEND_URL to your Vercel URL (see below).

Frontend (Vercel)

1. Go to Vercel -> Add New Project.
2. Select this repository.
3. Important: Set the Root Directory to frontend.
4. Add the following Rewrite in frontend/vercel.json (already configured in this repo) to point to your Render backend.
5. In frontend/app.js, ensure RENDER_API_URL points to your deployed Render URL.

---

🔗 Live Demo

Check out the live platform here: nexus-research-ai.vercel.app

---

---

🏗️ Architecture Guide

For a deep dive into the system architecture, justifications for technical choices, and interview Q&A, please refer to the Implementation Plan.

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.