DeushChat - A German Friend

Khizer Ansari — Sun, 04 Oct 2026 11:48:41 +0000

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

What I Built

I built DeushChat , An AI based German friend that helps you learn German , this not an tutorial based app , Its powered by open source AI and teaches the user as per scenarios , The User is asked for some basic details and chooses a scenario [like ordering mensa , University application , Opening a Bank account etc.] Based on the scenario the app asks user some questions and the user has to answer them , after this the user will be evaluated . This app runs without internet after downloading so it is easy to learn in place with bad network , My friend wants to pursue higher studies in Germany therefore , I made this app which will make him better at speaking German and handling many popular scenarios ! he will be well aware about his situation and know what he needs to say , DeushChat will help him throughout his German life!

Demo

deushchat.onrender.com

Code

ItsTheBegginer / DeushChat

DeuschChat 🇩🇪

A local, offline-capable German practice partner built for one student preparing to move to Germany. It runs entirely on your own laptop — no subscription, no cloud, no data leaving your machine.

The AI is powered by Ollama running a small open-weight Gemma model, so it works without an internet connection once set up.

Prerequisites

macOS (tested on macOS 13+)
Python 3.11 or newer
~2 GB of free disk space for the Gemma model

Install steps

1. Install Ollama and pull the model

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Start the Ollama server in the background
ollama serve &

# Download the model (~1.6 GB — only needed once)
ollama pull gemma2:2b

2. Get the project

# Unzip or clone into a folder of your choice, then enter it
cd /path/to/Deushchat

3. Install Python dependencies

pip3 install -r requirements.txt

4. Run the app

streamlit run

…

View on GitHub

How I Built It

Gemma 2B as the tutor. It's an open-weight model, which I run locally through Ollama, so inference happens entirely on his laptop.
Streamlit for the interface, served on localhost.
SQLite for his profile, conversation history, and flashcards.

Why Does Open Innovation Matter?

It works offline. Once the model is downloaded, nothing needs the internet. He can practice on a train or with bad wifi, and I tested a full conversation in airplane mode [add screenshot or recording].
His practice stays private. Language learners make embarrassing mistakes. Here they never leave his machine, and there is no account, no logging, and no company holding his conversations.
It costs nothing to run. No subscription and no per-token bill, which matters for a student on a budget. Practice can be unlimited without worrying about usage.

My Agent Session

Prize Categories

Best Use of Gemma: the whole tutor runs on Gemma, locally, through Ollama.
Best Use of Render : The project is deployed on Render

DEV Community: Khizer Ansari