I Built StudyNest to Make Scattered Study Notes Easier to Review

A developer built StudyNest, a browser-based Vietnamese study companion that stores notes in localStorage and uses the open-weight openai/gpt-oss-20b model via Groq's API to answer questions grounded in those notes, generate five practice questions with suggested answers, and produce short summaries. The prototype, created for the Hacktoberfest Weekend Challenge, moved inference to Groq after a local-model version required a download that did not fit the available disk space, and the server instructs the model to answer only from supplied notes. The developer tested it on a Python lesson note, where it correctly explained that append(10) changed [7, 8, 9] to [7, 8, 9, 10], summing to 34, and notes that exporting notes and interactive quizzing are planned.

StudyNest is a small Vietnamese study companion for someone reviewing scattered notes. You can save notes in the browser, ask a question grounded in them, generate five practice questions with suggested answers, or get a short summary. I built this during the Hacktoberfest Weekend Challenge. I tested it myself with a Python lesson note: when I asked why ket qua was 34, it explained that append 10 changed 7, 8, 9 to 7, 8, 9, 10 , whose sum is 34. I have not yet handed it to another person, so I cannot claim feedback from a friend. StudyNest uses the open-weight openai/gpt-oss-20b model through Groq's API. I can switch the model in a local configuration file without changing the study workflow. This also lets the app run without downloading a large model onto a computer with limited disk space. The app sends notes to Groq only when an AI feature is used; notes are stored in the browser's localStorage. The front end is one HTML file. A small Python HTTP server serves it and calls Groq through the official Python SDK. The server asks the model to answer only from the supplied notes and to say when the notes do not contain an answer. The API key stays in config.py , which Git ignores; config example.py shows the configuration format. My first local-model version required a download that did not fit the available disk space. Moving inference to Groq made the prototype usable on that machine, while keeping an open-weight model at its core. A future version should support exporting notes and let a learner answer quiz questions interactively.