SLEEPYHEAD

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

What I Built

I built Sleepyhead, a local-first AI routine companion designed for a friend who struggles with maintaining a consistent sleep and study routine.

Instead of building another generic sleep tracker, I wanted to make something that feels familiar to a developer: GitHub-style contributions, challenges, streaks, XP, badges, and progress tracking, but focused on building a healthier and more consistent daily routine.

Sleepyhead helps the user:

• Track sleep, wake-up times, study sessions, and daily routines
• Complete small routine challenges
• Build consistency through streaks, XP, levels, and badges
• See their routine history through a GitHub-inspired contribution graph
• Prepare for upcoming exams with realistic study plans
• Get personalized insights based on their actual routine data
• Talk to an AI coach about study planning and bedtime decisions

The goal isn't to tell someone to suddenly become a perfect 5 AM person. It's about helping them make small, realistic improvements that they can actually maintain.

I built it for a friend because this was a problem I could see in real life, rather than something I invented just to make an AI demo.

Demo

Live Demo:

https://dev-challenge1-sleepyhead-navy.vercel.app/

Code

GitHub Repository:

https://github.com/k-tanuj/DevChallenge1-Sleepyhead

The repository contains the frontend, backend, local AI integration, data models, and setup instructions.

*How I Built It
*
Sleepyhead is built as a full-stack application with a local-first architecture.

*Tech Stack
*

• Frontend: React, Vite, TypeScript, Tailwind CSS
• Backend: FastAPI, Python
• Database: SQLite
• AI: Ollama with an open-weight instruction model
• Charts & Visualization: Recharts
• Local Development: Ollama for local AI inference

The AI isn't just a chatbot sitting on top of the application.

Sleepyhead passes the user's actual routine data, study history, challenges, and upcoming exams to the AI so it can generate useful recommendations based on their current situation.

For example, a user can say:

"I have an exam Friday and three chapters left."

Instead of simply responding with generic study advice, the AI can look at the available time, previous study behavior, and bedtime target and create a realistic plan.

It can also answer questions such as:

"Should I study another hour tonight?"

The goal is to balance study progress with maintaining a sustainable routine.

The application also generates insights from the user's historical data, such as identifying study periods where they are consistently more productive or noticing when late-night studying starts affecting their bedtime consistency.

Gamification

I used a developer/community-inspired model for the gamification:

Personal Routine → Repository

Challenges → Issues

Healthy Actions → Contributions

AI Coach → Maintainer

Routine Profile → README

Achievements → Badges

The contribution graph gives the user a visual history of their consistency, similar to the contribution graph developers already understand.

Importantly, the system doesn't reward staying awake longer or studying until unhealthy hours. The focus is on consistency and sustainable progress.

Why Does Open Innovation Matter?

Open innovation made it possible to build Sleepyhead around an AI system that can run locally instead of depending on a proprietary cloud API.

Using an open-weight model through Ollama means the core AI experience can run on the user's own machine.

That matters for a personal routine application because the data involved can be quite private: sleep patterns, study habits, exam schedules, and daily routines.

With local inference:

• Personal routine data can stay on the user's machine
• There is no mandatory paid AI API
• The model can be changed or experimented with
• Developers can inspect and modify the AI pipeline
• The project remains accessible for experimentation and contribution

Open AI ecosystems also make projects like this more hackable. Instead of treating the model as a black-box API, I could design the application around the model and experiment with how real user data is passed into the AI.

For me, that's the biggest value of open innovation here: the AI becomes something you can build with, inspect, modify, and run yourself.

Prize Categories

• Build for a Friend
• Open Source / Open Innovation
• AI / Open-Weight AI

Thanks for checking out Sleepyhead!