CodePrep helps you practice coding interviews in a way that actually feels useful. Instead of grinding random problems, you can talk through solutions, write code, get tested, ask for hints, and receive clear feedback on what to improve. It includes voice mode, built-in code execution, study mode, and progress tracking, so you can practice like a real interview and get better faster.
Built this for a few friends who were prepping for coding interviews and kept saying the same thing: most prep tools help you solve problems, but not actually practice interviewing.
So I made CodePrep to feel a lot closer to the real thing. You can talk through your solution, write code, get tested, ask for hints, and get feedback that’s actually useful afterward.
The fun part: this was built entirely with AI agents. That made it possible to move fast, experiment a lot, and turn an idea that started as “this would be useful for my friends” into a real product.
Would love to hear what you think:
does this feel more useful than the usual coding prep flow?
what would make you actually come back and use it regularly?
The gap between "knowing how to solve problems" and "being able to explain your thinking out loud while solving them" is huge and most prep tools completely ignore it. Talking through your solution is exactly what interviews test and it's the hardest thing to practice alone. Smart angle. Congrats on shipping 🚀
This product literally doesn't even work. Completely bug riddled across so many cases, the voice interviewer feels so unnatural, and it burns through credits at an unbelievably fast rate (8 requests so far have cost me $0.22). Pure vibecoded slop that was evidently not even tested.
Cool idea. Suggestion/question: could it search the web and find suitable interview code questions to prepare on, for the particular company someone is aiming to join?
Congratulations on the launch! It's a very cool idea. I'm curious why you went for just OpenAI rather than e.g. supporting ollama or custom endpoints to make it fully local? Also, in your FAQ about "why Python only" it implies that actual python code is run, not just assessed: Is that just for user code or does the application write and run arbitrary code?
Amrutha Gujjar