Study Buddy: an offline GATE study partner for my friend, built with Ollama
This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend What I Built Study Buddy is an offline exam study partner I built for my friend Anushka , who is preparing for GATE . GATE has a huge syllabus, and she wanted quick answers and practice questions from her own notes. I also wanted her study material to stay on her laptop and not go to someone else's server. You add…
This project is an offline GATE study partner called Study Buddy, created for a friend named Anushka who is preparing for GATE. GATE has a large syllabus, and Anushka needed a quick way to access and review her own notes. Study Buddy was built to run entirely on her laptop, without uploading her notes to any external servers.
To create Study Buddy, I added Anushka's notes (in PDF, text, or markdown) to a folder. The program then processes the notes into chunks and assigns them local embeddings using an open-weight model called nomic-embed-text, which is run with Ollama on her laptop. When Anushka asks a question, Study Buddy fetches the most relevant chunks from the local database, and then uses another open-weight model, Qwen2.5:7b, to generate an answer that is cited with a source reference like [notes.pdf p.3].
If the notes don't have an answer for the question, Study Buddy politely indicates that the information isn't available instead of guessing.
In addition to answering questions, Study Buddy also quizzes Anushka on her own material by generating multiple-choice questions from the notes and explaining the correct answer along with the source. It keeps track of which topics Anushka struggles with, so she can focus on practicing those weak areas. Because everything runs on Anushka's laptop, she can use Study Buddy even without an internet connection.
The project is open-source and built using Ollama, Qwen2.5:7b, Nomic-embed-text, ChromaDB, SQLite, FastAPI, and plain HTML. The code can be easily run on any laptop with similar specifications, such as a 16 GB RAM Intel Core i7-13620H processor without a dedicated GPU. The first answer after starting the server takes about 55 seconds, while normal answers and quiz questions take approximately 42 seconds and 35 seconds, respectively.
Although slower than a cloud-based chatbot, Study Buddy still works effectively on an ordinary laptop.
One limitation of this project is that the 7B model may still misinterpret passages, so every answer includes a source reference to allow for verification. Additionally, the notes are currently added through a folder and an indexing command, rather than a direct upload button. Despite these limitations, Study Buddy provides privacy, cost-effectiveness, control over the models, and reproducibility for anyone with a similar laptop setup.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.