slide 1
slide 2
slide 3
slide 4
slide 5
slide 6
slide 7

Project Story: ExamCraft

About the Project

Inspiration

The inspiration behind ExamCraft came from the stress and inefficiency often associated with traditional exam preparation. I noticed that students spend countless hours sifting through past papers, trying to identify patterns and key topics. I wanted to create a tool that could streamline this process, using AI to analyze exam papers and generate personalized study materials. The idea was to help students study smarter, not harder, by focusing on the most relevant content.

What I Learned

Throughout this project, I learned a great deal about the power of AI and machine learning in education. I explored Natural Language Processing (NLP) with tools like spaCy and Hugging Face Transformers to analyze text and extract meaningful insights from past exam papers. I also gained experience in integrating machine learning models, like those from Google GenerativeAI, into a practical application that can genuinely assist users. Additionally, I deepened my knowledge of messaging platform integration, specifically with Twilio’s WhatsApp API.

How I Built the Project

ExamCraft was built using a combination of technologies and methodologies:

Backend: I used Django to handle user requests, manage the database, and serve the web application.
AI & NLP: For analyzing past papers and generating study materials, I utilized spaCy, Hugging Face Transformers, and Google GenerativeAI.
Messaging Integration: I integrated Twilio’s WhatsApp API to deliver personalized study materials directly to users’ phones.
User Interaction: The platform supports various input methods, including voice notes, handwritten images, audio files, and translations, providing a seamless and inclusive user experience.
AI Assistance: For deeper understanding, I implemented an AI call feature that provides users with real-time explanations on complex topics.

Challenges Faced

Building ExamCraft wasn’t without its challenges:

NLP Complexity: Understanding and implementing NLP to accurately analyze exam papers and extract useful data was a significant challenge. Fine-tuning the models to ensure they delivered relevant and accurate study materials required extensive testing and iteration.
WhatsApp Integration: Integrating Twilio’s API for WhatsApp posed its own challenges, especially in ensuring that messages were delivered reliably and formatted correctly across different devices.
User Input Flexibility: Allowing users to input questions in various formats (voice, text, images) required careful handling to ensure that the AI could process and respond accurately.
AI Clarity: Ensuring that the AI could provide clear and accurate responses during calls was another challenge, particularly in making the explanations understandable and contextually relevant.

Extra Features

Interactive Quizzes: To reinforce learning and provide immediate feedback.
Progress Tracking: Helps users monitor their study progress and identify areas for improvement.
Adaptive Learning Pathways: Adjusts content based on user performance.
Multilingual Support: Supports multiple languages for non-native English speakers.
Accessibility Options: Ensures the platform is accessible to all users.