🧠 JCAP AI Paper Summarizer

An AI-powered web application that summarizes research papers using Ollama and local LLMs like Mistral.

Supports:

• ✅ Single PDF summarization
• ✅ Batch ZIP upload of multiple PDFs
• ✅ Parallelized chunk processing
• ✅ Web-based interface with velvet-blue theming

This is a frontend interface only. It requires a running Ollama backend.

---

🚀 Features

Feature	Description
🔍 PDF Parsing	Extracts and chunks full paper text
🧠 LLM Summarization	Sends chunked text to Ollama (Mistral)
📁 ZIP Upload	Summarize multiple PDFs at once
⚡ Parallel Processing	Threads summarization for speed
💾 Download	Save summary as .txt
🌈 Themed UI	Inspired by Velvet Room / Persona aesthetic

---

---

🖼️ Screenshots

🧠 Main Interface

Upload a single PDF or a batch of papers as a .zip:

---

📄 Batch Summary Results

Each PDF is summarized in its own tab with section breakdowns:

📦 Project Structure

JCAP_AI_Paper_Summarizer/ ├── app.py # Flask app ├── requirements.txt ├── Dockerfile # Dockerfile for frontend (Flask) ├── ollama-backend/ │ └── Dockerfile # Dockerfile to host Ollama server ├── templates/ # HTML templates ├── static/ # Custom CSS styles │ └── style.css

---

📂 Requirements

• Python 3.11+
• Ollama running and serving models
• Models: mistral, llama3, etc.

---

🔧 Local Dev Setup

1. Create virtual environment

in bash run python3 -m venv venv source venv/bin/activate

2. Install dependencies

pip install -r requirements.txt

3. Start Flask server python3 app.py App runs at: http://localhost:8080

☁️ Cloud Deployment ✅ Frontend: Flask via Cloud Run Make sure Ollama is running and publicly accessible.

Build and deploy:

gcloud run deploy jcap-summarizer
--source .
--platform managed
--region us-central1
--allow-unauthenticated

Cloud Run will build from the included Dockerfile. 🧠 Ollama Backend (Required) This app requires an external Ollama backend running at:

http://<YOUR_OLLAMA_HOST>:11434/api/generate ✳️ Option A: Run Locally curl -fsSL https://ollama.com/install.sh | sh ollama serve ollama pull mistral ✳️ Option B: Run in Docker (Remote or Local)

cd ollama-backend docker build -t ollama-server . docker run -d -p 11434:11434 ollama-server Ollama must be running before deploying the Flask app.

🔒 Production Notes Ollama backend must be secured if deployed publicly (add HTTPS/auth/reverse proxy).

Cloud Run app expects Ollama at a static IP or domain.

For professional use, we recommend deploying Ollama on:

A GCE VM

A private VPS

Your own server

📄 License MIT