Year — 2026
SMARTSTUDENT-RAG
The Smart Study Recommender System is a Retrieval-Augmented Generation (RAG) tool designed specifically for academic use. Built to solve the "information overload" students face during exam seasons, this system ingests course materials (like lecture notes and past papers) and allows users to ask questions grounded strictly in their study data.
Stack
✨LLM: Groq (for high-speed inference)✨Vector Database: ChromaDB (Persistent local mode)✨Embeddings: Sentence Transformers (all-MiniLM-L6-v2 or BGE)✨Language: Python and Javascript
Key Features
- Adaptive Learning Modes:
- 🧑🏫 Tutor Mode: Detailed explanations with real-world analogies for grasping beginner concepts.
- 📝 Revision Mode: Concise, high-density summaries perfect for quick review sessions.
- 🎯 Exam Mode: Formats responses to mimic exam-style answers based on ingested past papers.
- 🧠 Conversational Memory: Remembers the session context (topics and subtopics), allowing for natural follow-up questions without repeating context.
- 🔍 Smart Content Recommendations: Uses metadata to suggest semantically similar content, topic adjacency, and learning progressions.
- 🛡️ Insufficiency Detection: If the system detects that a query is not covered in the notes (similarity score < threshold), it halts the RAG process and recommends external curated academic resources.
- 🔎 Full Traceability: A robust CLI and logging system records similarity scores, chunk IDs, and prompt execution for complete explainability.
- 🛠️ Tech Stack
Technical Highlights
- Context-Grounded QA: Answers are strictly generated from ingested academic content (PDFs, Markdown, DOCX) to ensure factual accuracy.
My Role
- Implemented the project fully