ThinkTutor

Login/Register Page
Homepage (Landing Page)
Chat interface (Offline) with metrics dashboard
Chat interface (Online)

Inspiration

Living in Lebanon and working with refugees and rural communities, I’ve seen firsthand the obstacles students face in accessing quality education — especially in regions with poor internet infrastructure. This experience, combined with my current role in the EdTech industry as a software developer, inspired me to build a solution that leverages technology while remaining accessible to all.

ThinkTutor was born out of a desire to use AI and edge-ready design to create an inclusive tutoring platform that works with or without reliable internet — empowering students in under-connected areas to learn without barriers.

What it does

ThinkTutor is an AI-powered tutoring platform that enables students (ages 12–15) to:

Ask educational questions via a chat interface.
Get real-time answers using AWS Bedrock generative AI.
Use the app offline — questions are saved and synced when back online.
Track key learning metrics: number of questions, sync status, up-time, and response time.

- Log in with a user account for a personalized experience.

How we built it

Backend: Django (Python) for user authentication, question handling, and AI integration.
Frontend: HTML, Bootstrap 5, and vanilla JavaScript to power the chat UI, offline logic, and metric tracking.
AI Integration: Connected to AWS Bedrock (amazon.titan-text-express-v1) via a secure client in bedrock_client.py.
Offline Capability: Used browser localStorage to queue questions and sync responses post-connection.

- Deployment: Hosted on AWS Lightsail, with future enhancements planned for AWS Snowcone deployment for true edge support.

Challenges we ran into

AWS IAM Permissions: Granting Bedrock access from Lightsail required custom IAM configuration not enabled by default.
Offline Sync Logic: Designing a seamless experience across offline and online states using only front-end tools.
Ensuring AI Safety: Tailoring AI prompts to be age-appropriate and informative.

- Real-time metrics: Implementing client-side metrics with persistence and reset capabilities.

Accomplishments that we're proud of

Created a fully functional AI tutoring system with offline-first architecture.
Integrated AWS Bedrock securely into a Django app.
Designed and deployed a clean, student-friendly UI.
Enabled learning insights with real-time metrics.

- Planned future deployment to Snowcone, extending reach to truly remote areas.

What we learned

Practical AWS Bedrock integration with server-based apps.
Best practices for designing offline-first educational tools.
How to handle IAM roles and policies in non-default AWS services like Lightsail.