Cache Overflow

Everyone knows how the average hacker learns from experience: trying, failing, forgetting, and trying again. Traditional notes apps let you record information, but they don’t reflect how your brain forgets over time. We wanted to build a system that actively tracks learning patterns, surfaces past mistakes, and helps prevent repeated errors.

The inspiration came from the workflow of many coders: how often we'd hit a bug we'd solved before, or forget an API nuance we thought we had mastered. We realized a “smart memory” for hackers could turn past mistakes into future insights, making learning more efficient.

Cache Overflow is a smart memory system for hackers. It tracks not just what you’ve learned, but how you've learned. By creating entries for notes, LeetCode questions, and previously encountered bugs, our AI system carefully tracks any patterns. Through our chatbox feature, users can then ask questions, and our AI answers based on past patterns if they're detected. This generates insights such as “Which mistakes cost me the most time?” or “What knowledge am I underusing?", helping hackers learn faster and avoid repeating the same pitfalls.

For this project, we used FastAPI for the backend, creating a system that logs entries into MongoDB. We incorporated Gemini with natural language processing to get relevant answers, including if questions reflect a repeating pattern or an already encountered solution.

Throughout this process, we ran into some challenges such as working in a virtual team and navigating version control issues while working asynchronously. We also ran into issues working with a free version of Gemini, which limited usage to 20 messages. Another challenge we faced was actually deploying the site due to handshake issues with MongoDB.

But, through this project, we learned how to use MongoDB and Gemini AI, which we were both first-time users of. We also learned how to communicate and work collaboratively, even when we weren't physically together. Finally, we learned how to better navigate React and API principles.

With more time, we hope to figure out how to deploy Cache Overflow in a way that is compatible with MongoDB. We also hope to add a user login/authentication system.

Built With

• fastapi
• geminiai
• mongodb
• python
• react