PrivaNote AI: A 100% Private, Local Study Tool Built with Gemma 2B

Mohisha Gupta — Mon, 05 Oct 2026 06:58:30 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

As an undergrad navigating heavy computing and math lectures, keeping notes organized is tough. I built PrivaNote AI for a close friend who constantly struggles to turn their chaotic, rapid-fire class notes into useful revision material before exams.
While they could just paste their notes into a cloud AI, they were uncomfortable handing over their personal academic drafts and lecture materials to a proprietary server. PrivaNote AI solves this problem. It takes their messy notes and instantly generates a clean summary, a bulleted list of key ideas, and a 3-question multiple-choice quiz—all without a single word ever leaving their laptop.

Demo

accounts.google.com

Code

mohisha1205 / privanote-ai

PrivaNote AI

PrivaNote AI is a privacy-first, locally hosted AI study tool built for the Hacktoberfest 2026 "Build for a Friend" Weekend Challenge.

It transforms messy, unstructured class notes into clear summaries, key concepts, and custom multiple-choice quizzes, helping students study more effectively without compromising their data privacy.

How It Works

PrivaNote AI uses an open-weight AI model that runs locally on your machine, keeping your academic notes on your device rather than sending them to a proprietary cloud AI service.

Frontend: A lightweight, interactive web interface built with Streamlit.
AI Engine: Google's open-weight Gemma 2B model.
Local Inference: Ollama runs the model directly on your hardware.
Privacy First: Your notes are processed locally, with no external AI API required.
Cost-Effective: No paid AI API calls are needed.

Prerequisites

Before running PrivaNote AI, ensure you have the following installed:

Python 3.8 or later - Download Python
Ollama -…

View on GitHub

How I Built It

The project is built entirely around open-source AI and local inference:

The AI Core: I used Google's open-weight Gemma 2B model.
Local Inference: I used Ollama to run the model directly on my local hardware.
The Interface: I built the frontend using Streamlit in Python, which takes the text input, wraps it in a strict system prompt, and sends it to the local Gemma model to generate the summary and quiz.

Why Does Open Innovation Matter?

Open innovation is the entire reason this project is possible. If I had used a closed API, my friend's personal data would have to be sent to an external server.
By using an open-weight model like Gemma running locally, I can guarantee 100% data privacy. It also means there are no API costs, no subscription fees, and the tool works completely offline. A student can use this in a basement library with zero Wi-Fi, and it will still work perfectly. Open AI makes powerful tools accessible and safe for everyday students.

Prize Categories

Best Use of Gemma

DEV Community: Mohisha Gupta