Inspiration
As students preparing for technical interviews and building projects, we’ve all faced moments when our classes provided the theory but not the practical guidance we needed. Questions like “Where is this file connected?” or “Why is my data structure implementation breaking?” can slow down progress and leave us scouring forums for answers. CodeMaven was built to solve this problem, acting as a mentor that’s always by your side to guide you through coding challenges and help accelerate your growth.
What it does
CodeMaven is an AI-powered coding extension designed to meet all your development needs. With its built-in chat interface, you can ask questions about your CS projects, and its multiple large language models (LLMs) will provide the best tailored response. The highlight-text feature allows users to select problematic code, receiving step-by-step explanations and optimization suggestions. CodeMaven doesn’t just fix errors—it teaches you the “why” behind each correction. With its voice-to-text capabilities, it’s like having a mentor who listens and responds to your every coding need, making it the ultimate personalized learning experience.
How we built it
LLM Integration: CodeMaven features multiple large language models, allowing users to choose their preferred AI tutor based on their coding challenges.
IDE Environment: We created a fully functional virtual IDE using a Python-based Judge0 IDE API, supporting both beginners and experienced developers.
Tech Stack: The frontend was built using HTML, CSS, and JavaScript, while the backend handles OpenAI’s API requests and user interactions.
Deployment: Leveraging cost-efficient deployment solutions like GoDaddy, we ensured easy access and scalability.
Challenges we ran into
One of the main challenges was optimizing the use of multiple LLMs while keeping the extension free for users. We implemented Open Router to access free tokens and seamlessly switch between different LLMs, ensuring that customers can experience the best AI coders available at no cost. Another major hurdle was integrating the front-end with the text-to-speech feature while ensuring smooth interaction between the user and the AI responses. Additionally, testing and training the AI coder to return relevant, context-specific feedback posed a challenge, requiring iterative tuning and fine-tuning prompts for various coding scenarios. Despite the time crunch, we managed to create a working model that provides tailored feedback and optimization suggestions effectively.
Accomplishments that we're proud of
User-Centric Design: We focused on building a tool that’s free and accessible to users of all levels. By finding cost-effective ways to deploy and optimize token usage, CodeMaven is designed to minimize barriers for students and developers.
Scalable Architecture: The platform is built to scale as user demand grows, making it adaptable for larger audiences over time.
Comprehensive Learning Tool: We’re proud that CodeMaven doesn’t just solve errors—it teaches users the underlying concepts and best practices, fostering long-term coding improvement.
What we learned
Balancing Functionality and Time: We learned how to prioritize features effectively to build a working prototype within a short timeframe.
Optimizing LLM Responses: Selecting the right models and training them to deliver context-aware, educational responses was a key learning curve.
Deployment Efficiency: Finding and using cost-effective deployment strategies taught us how to maximize value for end users.
What's next for CodeMaven
Expanded Language Support: We plan to add support for more programming languages and extend voice-to-text capabilities.
Collaborative Features: Future updates will include real-time collaborative coding for teams and pair programming.
Learning Analytics: Implementing user progress tracking and personalized learning paths to provide targeted coding exercises and tutorials.
Community-Driven Growth: Encouraging users to contribute feedback, suggestions, and improvements through an open-source approach.
Log in or sign up for Devpost to join the conversation.