PaperMind – Your AI Research Assistant
Inspiration
As students and researchers, we often find ourselves overwhelmed by lengthy research papers and the need to extract key insights quickly. We wanted to build something that simplifies the process — a smart AI assistant that can analyze, summarize, extract citations and insights, and even suggest similar research papers. Thus, PaperMind was born.
What It Does
PaperMind is an AI-powered web app that helps users:
- Upload research papers (PDF)
- Automatically extract the full text
- Summarize the paper and extract key insights
- Generate citations in APA format
- Extract named entities using NLP
- Recommend similar research papers based on embeddings
How I Built It
We used a modern full-stack approach:
Backend:
- Python + Django for API endpoints
- LangChain as the orchestration layer for toolchains
- Google Gemini API for summarization, insights, and citation generation
- spaCy for Named Entity Recognition (NER)
- FAISS for semantic similarity search
- PostgreSQL to store research metadata and embeddings
Frontend:
- Next.js + Tailwind CSS for a responsive UI
- Axios for file uploads and API requests
- Displays summaries, named entities, insights, and recommended papers
What I Learned
- How to integrate LangChain tools and build custom toolchains
- How to orchestrate an AI agent using Google's Gemini API
- Using FAISS with Django models for vector similarity search
- Efficient handling of PDF files in Django
- Connecting frontend and backend seamlessly for AI interactions
Challenges We Faced
- Handling large PDF files and converting them reliably into raw text
- Integrating LangChain's dynamic tools while working with an external LLM (Gemini)
- Getting FAISS to work smoothly with Django ORM
- Managing embeddings and aligning them with actual research paper entries
- Designing a clean UI that makes complex AI outputs easy to understand
What's Next?
- Adding support for more citation styles
- Integrating ArXiv or Semantic Scholar APIs to auto-fetch papers
- Real-time chat with the uploaded research paper using RAG
- Deploying the app on Vercel + Railway for public use
Log in or sign up for Devpost to join the conversation.