Skip to content
A.T.L.A.S. -Teaching Assistant

A.T.L.A.S. -Teaching Assistant

Ask questions. Think different. Learn together.

Created on 17th September 2024

A.T.L.A.S. -Teaching Assistant

A.T.L.A.S. -Teaching Assistant

Ask questions. Think different. Learn together.

Describe your project

The solution focuses on teaching Data Structures and Algorithms (DSA) using the Socratic method. It incorporates a Gen AI model that asks probing questions to help students understand key concepts in DSA. The assistant facilitates one-on-one learning, providing real-time guidance, answering questions, and assisting with code generation and debugging, specifically within the context of sorting algorithms. The AI adapts its responses based on the student’s input, leading to a personalized learning experience.

Our approach emphasizes guiding students through DSA using the Socratic method, which fosters deeper learning by encouraging students to explore concepts through interaction rather than receiving direct answers. While the assistant covers multiple DSA topics, the initial implementation prioritizes core areas within this domain, focusing on clarity and depth.

Currently, non-technical subjects, interactive coding environments, or full-course materials on DSA are out of scope. Additionally, the assistant is designed for one-on-one interactions, with no support for collaborative or group learning at this time.

Looking ahead, the solution presents several exciting possibilities. Recent enhancements include a notes feature that allows students to summarize learning points and a performance analysis feature that evaluates a student's progress based on interactions. Future developments could further enrich the learning experience with quizzes to assess comprehension and the expansion of a web version for broader accessibility. These features aim to make the assistant more interactive and adaptive to each user’s needs.

Challenges we ran into

The challenges we ran into during the development of this project were multi-faceted and required careful decision-making. One significant challenge was maintaining Git version control, particularly while working with both backend and frontend components in the same repository. Managing branches, resolving merge conflicts, and ensuring smooth collaboration across team members required meticulous attention to detail, especially when changes were happening simultaneously on both the backend and frontend. Additionally, keeping the repository organized and ensuring proper version control for each module was a crucial but challenging task that we had to overcome.

Another challenge was choosing the right frontend technology. We initially debated between React.js and Flutter, as both offered distinct advantages. React.js is widely used for web-based interfaces, while Flutter offers cross-platform support and would allow us to target both web and mobile with a single codebase. After much deliberation, we chose Flutter for its flexibility and ease of integrating the AI-powered teaching assistant on both mobile and web platforms. However, this decision brought its own learning curve and added complexity in setting up the environment, which we had to navigate effectively.

Lastly, model training presented its own challenges. Training the AI model to not only understand sorting algorithms but also respond Socratically required tuning the responses carefully to make them feel more natural and engaging for students. Achieving the right balance between guiding the students and providing useful hints, without directly giving away answers, involved multiple iterations. Fine-tuning the model and ensuring it could handle various queries in real-time took considerable time and effort.

Tracks Applied (1)

5. Problem statement shared by Blume Ventures

Our solution addresses the challenge by creating a Gen AI-powered teaching assistant focused on Data Structures and Algo...Read More

Discussion

Builders also viewed

See more projects on Devfolio