Ai Sage

Our Hackathon Journey: Building an AI-Powered Student Assistant Inspiration: Addressing Student Struggles with Information Overload As students step into a new academic environment, they often find themselves overwhelmed by scattered and unstructured information across various websites, portals, and social media platforms. Whether it’s finding immigration policies, class schedules, professor details, or registering for events, the lack of a unified knowledge base creates frustration and inefficiency.

This problem resonated with us personally, as we have either faced it ourselves or seen friends struggle with it. We envisioned a solution that would consolidate all necessary information into a single, AI-powered assistant—an intuitive, personalized bot that simplifies access to relevant details, enables event registration, and provides notifications, all while offering a seamless, speech-based interaction experience.

How We Built Our Project: Tech Stack & Development Process We leveraged a combination of AWS cloud services, generative AI, and a modern web application framework to create our AI-powered student assistant. Here’s how we structured our solution:

Data Collection & Processing:

We used Amazon S3 to store structured and unstructured student-related data, including event schedules, course syllabi, and professor details.

Web Scraping & OpenSearch: To gather information from multiple sources, we implemented web scraping techniques and indexed the extracted data into Amazon OpenSearch for fast retrieval.

We created embeddings from this data and stored them in a Knowledge Base using Amazon Bedrock to enhance AI-driven responses.

Backend & AI Model Integration:

We implemented AWS Lambda to handle API requests and execute AI-driven actions dynamically.

Amazon Bedrock Agents powered the chatbot with foundation models to generate personalized responses based on student queries.

The AI model was fine-tuned to retrieve relevant information from the Knowledge Base, ensuring students received accurate and contextual answers.

Frontend & User Interaction:

We built a React Single Page Application (SPA) to provide a user-friendly interface.

The UI was hosted and accelerated via Amazon CloudFront for better performance.

Students could interact through a chat-based interface, which connected to the AI via API Gateway.

We incorporated speech-based collaboration to enable students to use voice commands instead of just text input.

Personalization & Notifications:

The bot recommended events based on user preferences and academic schedules.

Users could automatically register for events, and confirmation emails were sent via AWS email integration services.

Amazon Cognito handled authentication and authorization, ensuring secure access.

Built With

• amazon-bedrock
• amazon-dynamodb
• amazon-web-services