✦ LUMA — PRIVATE AI STUDY COMPANION

hf26challenge

WHAT I BUILT

I built Luma, a private AI study companion designed to help students study more easily while keeping their study material local.

Luma has four main features:

• Explain difficult topics in simple language
• Generate beginner-friendly quizzes
• Simplify complicated study material
• Upload PDF notes and ask questions about them

The AI runs locally through Ollama using the open-weight Llama 3.2 model, so study questions and uploaded notes can be processed on the user's own computer instead of being sent to a closed cloud AI API.

I built Luma to solve a real student problem: getting useful AI assistance for studying while keeping personal study notes and educational material under the user's control.

DEMO VIDEO:

SOURCE CODE:

Student
   ↓

SHOW THE CODE

Student
   ↓

The main application is:

app/web.py

The project also contains:

• README.md
• requirements.txt
• .gitignore

The application code connects the Streamlit interface to the locally running Ollama API, while PyPDF is used to extract text from uploaded PDF notes.

HOW I BUILT IT

I built Luma using Python and Streamlit for the web interface.

Ollama provides local AI inference using the open-weight Llama 3.2 model. Luma communicates with Ollama's local API to generate explanations, quizzes, simplified text, and answers.

For PDF notes, I used PyPDF to extract text from uploaded documents before using the local Llama model to answer questions about the extracted content.

Technology stack:

• Python
• Streamlit
• Ollama
• Llama 3.2
• PyPDF
• Local HTTP API

The basic architecture is:

Student
↓
Luma Web Interface
↓
Python + Streamlit
↓
Ollama
↓
Llama 3.2
↓
AI Response

For PDF questions:

PDF Notes
↓
PyPDF
↓
Extracted Text
↓
Ollama + Llama 3.2
↓
Answer From Notes

WHY DOES OPEN INNOVATION MATTER?

Open innovation matters because it gives developers more control over the technology they build with.

For Luma, using an open-weight model with local inference made it possible to build an AI study companion without making a closed cloud AI API the core dependency.

This is especially useful for private study material. Students can run the AI locally, experiment with the model, customize the application, and build new features around an open technology stack.

Open innovation also makes experimentation and learning more accessible because developers can build around open technologies instead of treating AI as a completely closed service.

MY AGENT SESSION

I did not use a separate autonomous agent session for this project.

Luma is a local AI application built around Ollama and Llama 3.2 rather than a multi-step autonomous agent framework.

PRIZE CATEGORIES

GitHub

I am entering the partner category/categories that directly match the technologies used in Luma, particularly the Ollama/local AI category if available in the submission form.

Luma uses Ollama as the local inference layer and Llama 3.2 as its open-weight AI model.

TEAM SUBMISSION

This is a solo project.

I am the sole creator and submitter of Luma.

Dev username:
@c_k_74275275b328c851fce40