This week, we are launching codercards.ai, an application that analyzes a user’s GitHub activity to provide insights on their habits and practices. I am incredibly proud of the team for putting this together in a matter of weeks from conception to MVP, revision, and now a v1 release!
The project organically grew from our research into and ongoing development of ways to make LLM-generated code better align with user-generated code. @Roger_Wiebe demonstrated his initial progress into code analysis and the team was very impressed with the characteristics he was able to extract from a modest input sample. Pierce saw the opportunity to present some of these insights in a visual demo and had a proof-of-concept together in a matter of days. Once we committed to putting this project together, @charliebailey designed and implemented the data collection that enables our on-demand card generation. Check out our company announcement for more on the actual project implementation.
Every individual contributor brought ideas that shaped the end result we are sharing now; in a time where teams are shrinking, I think it is particularly important to highlight that a diverse set of perspectives and ideas are what made this project happen. While we are actively working on making LLM Code-Generation look more human, the goal was never to replace humans, we rather view our tools and technology as ways to assist the implementation of human ideas and innovations.
