InnoTutor Presents ChatEDU(Team:Ashfire)

⚜️ Drumroll, here...comes InnoTutor's ChatEDU

• InnoTutor's ChatEDU is a revolution in learning, not merely a teaching tool.
• We've gone beyond conventional learning techniques by utilising the capabilities of Azure AI, including GPT-4 and Azure's varied datasets and services.
• With the help of our platform, you can turn any text or file into an engaging tutoring session that includes context-driven learning, multiple-choice exams, extended answer formats, and personalised study guides.
• This is education reimagined paradigm shift on education.

Problem Statement

• In today's rapidly evolving educational landscape, traditional teaching methods often fall short in providing personalized and contextualized learning experiences. To address this challenge, InnoTutor has developed ChatEDU, a revolutionary educational platform powered by Azure AI technologies and OpenAI's GPT-4.
• However, despite its advanced capabilities, InnoTutor faces the need to ensure seamless content upload and extraction processes, optimize the Retrieval-Augmented Generation (RAG) prompting flow, and streamline the continuous integration and deployment (CI/CD) pipeline using GitHub Actions.
• As such, InnoTutor seeks to formulate a comprehensive solution that enhances data processing efficiency, improves generative language model performance, and automates deployment processes to ensure consistent delivery of the platform's latest features and updates.

💻 Technical Architecture

• This solution makes use of PostgreSQL for reliable database management, Microsoft Entra ID for secure authentication, Azure Cosmos DB for dynamic front-end development, and OpenAI's GPT-4 for intelligent data handling and conversational AI.
• InnoTutor offers a smooth learning experience from information retrieval to knowledge assessment because it was designed with scalability and user experience in mind.

Tech Stack

• Microsoft Entra ID(Azure AD)
• Next.js
• Typescript
• Azure AI Search
• Chakra UI
• Azure AI Document Intellignce
• Azure Cosmos DB

⚡ Content Upload and Extraction Flow

• Through a client interface, users can submit files or text to InnoTutor, and the data is then sent to the endpoints.
• After processing (text is taken out of files using Azure AI Document Intelligence), this data is saved in Azure Cosmos DB so that it can be used in other platforms.

UML Diagram

📜 Challenges we ran into

• Integration of GPT-4 & Azure Services: Integrating the latest AI advancements with Azure services was no easy task. It required meticulous planning and adaptation to ensure seamless harmony. We overcame this hurdle through iterative testing, and leveraging the extensive documentation provided by both OpenAI and Microsoft Azure.

• Data Processing and Storage Optimization: Processing and storing large amounts of data can be challenging, but with Azure AI Document Intelligence, it's a breeze! Speed and scalability are no longer an issue. We optimized our code, used parallel processing, and fine-tuned Azure Cosmos DB. The result? Improved performance and responsiveness!

• Synchronization of Search Service and Database: We tackled the challenge of keeping search results current by implementing real-time synchronization mechanisms. Now our search service always displays the latest information from the database!

📁 RAG Prompting Flow

• InnoTutor's Retrieval-Augmented Generation (RAG) prompting flow begins when a user provides a prompt via the client interface, which triggers a request to the '/api/chat' endpoint.
• The system then extracts relevant information from the Azure Cosmos DB database to give context for the generative language model (LLM), which uses the data to generate personalised educational content.
• Ultimately, the response is broadcast back to the client, giving the user an answer enhanced by the data that was retrieved, guaranteeing a more precise and contextualised interaction.

📤 Inside The Box(Azure AI Search)

• We index our database's users and notebooks collections using Azure AI search.
• As a result, users are able to search the data through the client quickly and effectively.
• The search results are constantly current since the search service and database are automatically synchronised.

🗂️ Continuous Integration and Deployment

• GitHub Actions is used to construct the CI/CD pipeline for InnoTutor. With each push or pull request to the repository, the code integration, testing, and deployment processes are automated.
• The application is deployed to Azure Static Web Apps after the automatic procedures have confirmed the modifications, guaranteeing continuous delivery of the most recent platform version.

Try it out

• https://azure.chatedu.io/

Github

• Github: https://github.com/akashpanda122/innotutor
• Demo Link: https://www.loom.com/share/b0c0c2268cf54bcbbe53f8995a91ef22?sid=5cb6d3ed-0d50-43e5-9b51-ffc717f82503

📜 Licence

InnoTutor is available under the MIT license. See the LICENSE file for more info.