Shiksha Miraz is designed to address several critical issues in the education and learning sector:
The Problem: Traditional education lacks adaptability, offering a one-size-fits-all approach.
Solution: Shiksha Miraz delivers personalized learning using AI and Large Language Models (LLMs) to cater to individual learning styles and paces.
The Problem: Unequal access to quality educational resources limits opportunities, especially in underserved areas.
Solution: Shiksha Miraz provides universal access to a vast repository of educational content, bridging the knowledge gap.
The Problem: Conventional assessment methods are time-consuming and subjective.
Solution: Shiksha Miraz streamlines assessments with automated question generation and real-time grading for instant feedback.
The Problem: Lack of transparency in educational progress and AI algorithms.
Solution: Shiksha Miraz emphasizes transparency, offering insights into student performance and AI-driven recommendations.
The Problem: Resource limitations hinder learning for students and educators.
Solution: Shiksha Miraz provides a wealth of educational resources, including summaries, questions, answers, and learning materials, leveling the playing field.
Shiksha Miraz aims to revolutionize education by enhancing accessibility, personalization, efficiency, and transparency for all users.
Our journey in developing Shiksha Miraz came with its fair share of challenges. We've overcome these hurdles through dedication, teamwork, and creative problem-solving:
Designing a visually appealing and user-friendly interface in Flutter was no small feat. We needed to create an engaging and intuitive platform for our users while ensuring a consistent and polished design throughout the application.
How We Overcame It:
Our design team worked tirelessly to craft a beautiful and responsive UI. We conducted user testing and gathered feedback to fine-tune our design, resulting in a user-centric experience that meets our high standards.
As the project evolved, we realized the need for a more efficient database structure and real-time data synchronization. This required a significant overhaul of our database system.
How We Overcame It:
We embraced the challenge by adopting a robust database management approach. With a combination of firebase database and real-time syncing solutions, we redesigned our database architecture for improved performance and data consistency.
Working with LLMs introduced us to unique challenges. Sometimes, these models exhibited hallucinations, and fine-tuning them for educational content required a substantial time investment. Ensuring both interactivity and performance while handling LLMs was another complex task.
How We Overcame It:
We followed the CRISP-DM (Cross-Industry Standard Process for Data Mining) approach, which helped us effectively fine-tune the LLMs for educational purposes. This iterative process allowed us to reduce hallucinations and optimize performance. Additionally, we enhanced our server infrastructure to handle the computational demands, resulting in a faster and scalable learning experience.
Through these challenges, we've gained valuable insights and expertise.
Tracks Applied (1)
Major League Hacking
Discussion