📚 Elyaitra – Syllabus-Bound AI Study Assistant

Elyaitra is an AI-powered study assistant designed specifically for students — but with a twist:
It only answers from the official syllabus. Nothing extra. Nothing irrelevant.

No note uploads. No random internet answers. Just exam-focused, syllabus-accurate help.

---

🚀 What is Elyaitra?

Elyaitra solves a very common student problem:

“This answer looks good, but is it in our syllabus?”

Elyaitra:

• ✅ Comes with preloaded syllabus & content
• 🎯 Answers strictly from exam syllabus
• ❌ Clearly says “Not in Syllabus” if a topic is not required
• 🧠 Prevents over-studying & irrelevant learning
• ⚡ Works instantly with zero setup

---

✨ Key Features

• 📚 Preloaded Syllabus, Zero Setup
  Students don’t upload notes or PDFs. Everything is already structured and indexed.

• 🎯 Strictly Exam-Bound AI
  The AI is restricted to only answer from the allowed syllabus content.

• ❗ Clear “Not in Syllabus” Responses
  If you ask something outside the syllabus, Elyaitra tells you directly.

• 🔍 Fast Semantic Search + AI Answering
  Uses vector search + AI to retrieve and generate accurate, relevant answers.

• 🧩 Subject → Unit → Topic Based Flow
  Fully structured and organized for real exam preparation.

---

🛠️ Tech Stack

Frontend

• ⚛️ React (Vite)
• 🎨 Tailwind CSS
• 🌐 Deployed on Vercel

Backend

• 🐍 Python (FastAPI)
• 🧠 Google Gemini API
• 🧲 Vector Database (ChromaDB)

AI / Retrieval

• 🔎 RAG (Retrieval Augmented Generation)
• 📦 Multi-LLM Provider Support (Gemini & Ollama)
• 📦 Embeddings for syllabus content
• 🗂️ Chunked & indexed syllabus data

---

⚙️ AI Provider Configuration

Elyaitra now supports switching between Google Gemini (default) and a locally hosted Ollama server.

Switching Providers

Update the backend/.env file:

# To use Gemini:
LLM_PROVIDER=gemini
GEMINI_API_KEY=your_key_here

# To use Ollama:
LLM_PROVIDER=ollama
OLLAMA_BASE_URL=http://localhost:11434

Models Used

• Gemini: gemini-2.5-flash / gemini-embedding-001
• Ollama: llama3.1:8b / nomic-embed-text

---

🧠 How It Works (Architecture)

1. Syllabus content is preloaded and embedded into a vector database.
2. User selects:
   • Subject
   • Unit
   • Topic (or asks a question)
3. The system:
   • Retrieves only relevant syllabus chunks
   • Sends them to the AI
   • Generates an answer only from that content
4. If nothing relevant is found:
   • ❌ Returns: “Not in Syllabus”

---

📸 Demo

(Add screenshots / demo video link here)

Chatbot--