Project Info

RetAIn

Inspiration

In today's fast-paced academic world, students and young professionals often find themselves overwhelmed by complex concepts and dense academic materials. Whether it's sifting through lecture notes, deciphering lengthy reports, or struggling to understand a difficult research paper, the challenge of grasping key ideas can feel daunting. This confusion can slow down the learning process, leaving students frustrated and unsure of where to focus their efforts. This struggle highlights the necessity for tools that can bridge the gap between overwhelming content and true understanding, offering a more efficient and accessible way to digest critical information. What RetAIn does RetAIn harnesses the power of AI to transform academic materials into personalized, easy-to-understand audio lectures. Whether you’re dealing with notes, research papers, or reports, RetAIn takes your content and breaks it down into tailored explanations, guiding you through the most important concepts. By identifying key themes and simplifying complex information, RetAIn turns your study materials into a structured learning experience, helping you focus on what really matters.

How we built it

We designed the project specifically for school students, aiming to help them streamline their learning process. Frontend We used React to create a clean and user-friendly interface. Backend Our backend stack consisted of Node.js and an Express server connected to a PostgreSQL database and Amazon S3, which managed user data and stored documents. AI Integration Our flask service implements a Retrieval-Augmented Generation (RAG) model using a Flask API. It processes PDF documents, generates embeddings using OpenAI's model, and stores them in a Pinecone vector database for efficient retrieval. When a question is asked, the system queries Pinecone to find relevant text chunks, then uses OpenAI's GPT-4 model within an AutoGen multi-agent framework to generate a comprehensive response. This RAG system enables intelligent processing of lecture materials, generating summaries, answering questions, and creating study guides based on uploaded documents. For text-to-speech generation, we leveraged the Cartesia TTS API to provide a clear and natural-sounding speaking experience.

Challenges we ran into

One of the biggest challenges we faced was downsizing the project scope. Initially, we wanted to implement highly advanced features, such as generating article-like outputs with animated visualizations. However, this idea proved too complex for our timeline and resources, as it would have required incorporating rendering engines and advanced animation technologies. After realizing the technical and time constraints, we shifted our focus to building a more streamlined application that handled note summarization and study guide generation effectively. It also became a difficulty incorporating sponsor material effectively, so we ended up falling back on our technical backgrounds more with the experience we had before coming here. Another challenge was finding a balance between simplicity and functionality. We had to make sure the AI-generated content was useful without overwhelming the user with too much information, which required fine-tuning the RAG AI model for better relevance and accuracy.

What we learned

Throughout this project, we gained valuable insights into both front-end and back-end development as well as the use of AI in practical applications. We learned how to effectively build an intuitive user interface, manage databases for large sets of documents, and integrate AI for real-time note summarization and study guide generation. We also gained experience in scaling back ambitious features in favor of a more focused and usable product, learning to prioritize simplicity and functionality. Conclusion Despite the hurdles, we successfully built a functional and user-friendly AI-powered note-taking app. The project not only provided a valuable learning experience in full-stack development and AI integration but also helped us understand the importance of prioritizing and adapting our goals to match our capabilities and timeline.

Retain - AI-Powered Lecture Generation Platform

Overview

Retain is an intelligent platform that transforms academic materials into personalized, easy-to-understand audio lectures using AI. It helps students and professionals tackle the challenge of processing complex academic content by converting documents into structured, accessible learning experiences.

Problem Statement

Students and professionals face several challenges with traditional learning materials:

Overwhelming volume of complex academic content
Difficulty in identifying and focusing on key concepts
Time-consuming process of digesting dense research papers and notes
Need for more accessible and efficient learning methods

Solution

Retain addresses these challenges by:

Converting academic materials into personalized audio lectures
Breaking down complex concepts into digestible explanations
Using AI to identify and focus on key themes
Providing a structured learning experience through an intuitive interface

Tech Stack

Frontend

React.js: Core frontend framework
Tailwind CSS: Utility-first CSS framework for styling
React Router: For navigation and routing
Lucide React: For UI icons and components
ShadCN UI: Component library for modern UI elements

Backend

Node.js & Express: Server framework
PostgreSQL: Primary database
AWS S3: Document storage
Multer: File upload handling
JWT & Bcrypt: Authentication and security
CORS: Cross-origin resource sharing

AI & Machine Learning

Flask API: RAG service implementation
OpenAI GPT-4: For text processing and generation
Pinecone: Vector database for embeddings
AutoGen: Multi-agent framework for AI processing
Cartesia TTS API: Text-to-speech conversion

DevOps & Infrastructure

Docker & Docker Compose: Containerization and orchestration
AWS Infrastructure: Cloud hosting and services
Poetry: Python dependency management
NPM: Node.js package management

Key Features

Document Processing

PDF document upload and processing
Intelligent text chunking and embedding
Secure document storage in AWS S3

AI-Powered Analysis

Retrieval-Augmented Generation (RAG) for context-aware responses
Multi-agent processing using AutoGen
Intelligent summarization and key concept extraction

User Experience

Clean, intuitive web interface
Personal course and lecture management
Audio lecture generation and playback
Progress tracking and management

Setup and Installation

Prerequisites

Node.js (>= 12.0.0)
Python 3.12
Docker and Docker Compose
AWS account credentials
PostgreSQL

Backend Setup

Clone the repository
Navigate to the backend directory:

cd backend

Set up PostgreSQL using Docker:

docker-compose up -d

Install dependencies:

npm install

Configure environment variables:

DB_USER=postgres
DB_HOST=localhost
DB_NAME=retain_db
DB_PASSWORD=your_password
DB_PORT=5432
AWS_ACCESS_KEY_ID=your_aws_key
AWS_SECRET_ACCESS_KEY=your_aws_secret
AWS_REGION=your_aws_region

Start the server:

npm run dev

Frontend Setup

Navigate to the frontend directory:

cd frontend

Install dependencies:

npm install

Start the development server:

npm start

RAG Service Setup

Navigate to the RAG service directory
Set up Python environment:

poetry install

Configure API keys:

OPENAI_API_KEY=your_key
PINECONE_API_KEY=your_key

Start the Flask server:

poetry run python agent.py

API Endpoints

Authentication

POST /signup: User registration
POST /login: User authentication

Courses

POST /create-course: Create a new course
GET /courses/:email: Get all courses for a user

Lectures

POST /create-lecture: Create a new lecture
GET /lectures/:email/:course_id: Get all lectures for a course

Contributing

We welcome contributions! Please read our contributing guidelines and submit pull requests for any improvements.

License

This project is licensed under the ISC License.

Acknowledgments

OpenAI for GPT models
All contributors and supporters of the project

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

CSSIn code
ExpressIn code
FlaskIn code
HTMLIn code
JavaScriptIn code
LangChainIn code
OpenAIIn code
PostgreSQLIn code
PythonIn code
ReactIn code
SQLIn code
Tailwind CSSIn code
AWSClaimed
DockerClaimed
Node.jsClaimed

12 of 15 appear in the indexed code. 3 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

76 KB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.