HuskyTrack

Inspiration

We were inspired by the need for fast, accessible academic advising, especially in last-minute situations. For example, if it’s 9 PM the night before course registration, meeting with a real academic advisor isn’t possible. That’s when we thought: what if we built an AI that could help?

To bring this idea to life, we built a web-based platform that includes a user dashboard and a chat-based AI assistant. The dashboard allows users to upload their transcripts and DARS reports, giving the system context about their academic progress. It also helps students track their degree progress over time in an easy-to-understand format.

What it Does

The chat feature enables users to interact with an intelligent agent that provides tailored course recommendations. It takes into account:

• Courses the user still needs to take
• Available classes based on the real-time teaching schedule
• Instructor information pulled from Rate My Professor, including ratings and reviews
• Peer feedback and common student experiences

To build the project, we divided the work among our team. One member developed the front-end UI and dashboard. Another handled web scraping for course schedules and Rate My Professor data. A third teammate managed user authentication, while another integrated and fine-tuned the AI using Amazon Bedrock.

Challenges

We faced significant challenges, particularly in handling user-uploaded data like transcripts and DARS reports. Parsing this information reliably was complex due to formatting inconsistencies. We also struggled with AI model versioning and training compatibility across different environments. However, through persistent debugging and thorough reading of documentation, we were able to get everything working together smoothly.

What we learned

As a team, we learned a lot, especially around tools that were new to us. This included working with AWS services, setting up and deploying using Amazon Bedrock, and integrating an AI-powered chat client into a real-world application.

Built With

• amazon-web-services
• bedrock
• claude
• express.js
• lamda
• python
• react.js
• s3