This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend My best friend, Laiba, is one of the smartest people I know, but she faces a massive hurdle every semester: studying is just plain boring to her. The moment she opens her computer science textbooks, her motivation flatlines. She gets easily overwhelmed by endless to-do lists, and traditional productivity apps only make her feel guilty when she falls behind. I wanted to build something that didn't just track her work, but actually loved her through the process.
So, I built StudyBuddy: a completely custom, hyper-gamified, kawaii-themed study dashboard designed specifically to make Laiba smile.
Instead of a cold checklist, every task she completes earns her XP and keeps a virtual Tamagotchi-style plant happy and bouncing. If she procrastinates for over 24 hours, the plant visually wilts and begs for water. But the absolute core of the project is the StudyBuddy AI—a local AI companion powered by Google's Gemma model that sits in the corner of her screen. Whenever she feels stressed or bored, she can rant to the AI, and it replies with hyper-supportive, sweet, and motivating micro-encouragements (always ending with a cute "♡").
When I finally showed it to her, she actually teared up. She said, "It feels like someone is holding my hand while I do my DSA problems." She hasn't broken her study streak since.
🌸 Live Demo:
https://youtu.be/aAaTiMc6UKQ A cute, gamified study companion that turns everyday studying into a small, rewarding experience. 🌱✨
StudyBuddy is a student-focused productivity web app designed to make studying feel less overwhelming and more engaging.
Instead of treating studying like a boring checklist, StudyBuddy turns completed tasks into XP, maintains a study streak, and keeps a virtual Tamagotchi-style plant growing alongside you.
At the heart of the experience is StudyBuddy AI, a local AI companion powered by Google's Gemma model that can provide motivation, study suggestions, and simple daily insights.
I built StudyBuddy as a full-stack web application with a heavy focus on local AI inference.
Frontend: React and Tailwind CSS, heavily customized with SVG assets, dynamic CSS filters (for the plant's mood states), and localStorage to persist Laiba's daily login streaks.
Backend: A lightweight Python Flask server that acts as the bridge to the AI.
The AI Core: I utilized Gemma 2B, Google's open-weight model, running entirely locally on my machine via Ollama. I configured the Flask backend to ping the local Ollama API (localhost:11434) using the requests library.
To make the experience feel snappy and human-like for Laiba, I implemented a keep_alive parameter in the API payload so the Gemma model stays loaded in system memory during her study sessions, eliminating cold-start delays when she needs instant motivation. I also set up a /quote route that asks Gemma to generate a fresh, unique motivational quote every single time she opens the dashboard.
The application is organized around a simple study loop:
Student → Study Tasks → XP → Progress → Plant → AI Companion
the study dashboard, task completion,
XP, streaks, plant state, and AI interaction.
For the AI component, I used Google's open-weight Gemma model as the StudyBuddy AI companion.
For StudyBuddy, open innovation wasn't just a technical choice; it was the only way this project could exist for Laiba. Radical Privacy for Vulnerability: When Laiba is stressed out about a failed exam or venting about burnout, she needs to feel safe. Because Gemma 2B runs 100% locally on her own hardware, her personal anxieties, study habits, and chat logs never leave her laptop. There is no corporate cloud reading her data.
Zero Cost for Broke Students: As college students, we cannot afford to pay monthly subscriptions or token-usage fees for API calls. Open-weight models like Gemma mean Laiba can talk to her StudyBuddy all night during finals week without me worrying about waking up to a $50 API bill.
Offline Reliability: Campus Wi-Fi is notoriously terrible. Because this is a local inference stack, Laiba can take her laptop to a quiet park with absolutely no internet connection and her AI companion will still be there to encourage her. Open AI models turn powerful technology into a local utility.
For StudyBuddy, using an open-weight model matters because the AI companion is meant to become part of a student's
personal study environment.
Using Gemma gives the project more flexibility over how the AI is run and integrated. The model can be experimented with,
adapted, and served in different environments instead of
making the entire application dependent on a single closed
AI API.
This is especially useful for a study companion because
students may want greater control over their data and the
ability to experiment with the AI locally.
Open models also make it easier for developers to understand,
modify, and build on top of the technology behind their
applications.
I didn't use DevRelay, but you can read my entire chat session with Gemini where we built the local Gemma integration together right here:
[View my Gemini Chat Session]https://share.gemini.google/BQf8bSC29Vbd Best Use of Gemma: I am running Google's open-weight gemma:2b model locally via Ollama to power both the interactive chat companion and the dynamic daily quote generation, serving as the motivational heart of the application.
StudyBuddy uses Google's open-weight Gemma model as the AI
companion behind the study experience.
Gemma is used to provide study-focused motivation,
suggestions, and progress insights rather than functioning
as a generic chatbot.