I’m definitely not a competitive programmer, I’m a product person at heart, but I enjoy the puzzle-solving side of our field. Most platforms focus on the "grind" of interview prep (certainly important) but I wanted to build something that felt more like a daily habit (think Duolingo) for sharpening your technical logic and how you explain it.

The Concept: It’s a collection of 400+ whiteboard-style challenges. Instead of just checking if your code runs, I’m using Gemini 3 Flash to grade you against a rubric. It looks at your approach, edge cases, and how you communicate trade-offs. I also made some cool "Peer Review" scenarios where you chat with a peer reviewing their code.

The goal isn't just to "solve it," but to keep that "whiteboard muscle" healthy. I also added some university-specific tracks (MIT 6.006, Stanford CS161) just because I thought they were a great way to revisit foundational logic.

The Tech: React 19, TanStack Router/Query, Supabase, and Tailwind.

I jumped on this now because Gemini 3 Flash is finally at a point where fast, cost-effective AI grading feels viable for a side project. Free accounts will come with ads to support (eventually).

Testing the waters: I’m still refining the AI feedback and the challenge difficulty, so I’m keeping the group small for now. I’ve opened spots for the HN community to skip the waitlist and get 6 months of premium for free as a thanks for trying it out, plus a cool HN Badge for your profile.

Link: https://dayleet.com Code: SHOWHN26 Sign up form: https://dayleet.com/signup?invite=SHOWHN26

I’d love to hear your thoughts on the quality of the AI grading. Does it actually help you think through the problem, or does it feel like a gimmick? I'm building this to help myself get better at algorithmic thinking this year as AI can do more of the heavy lifting with code, so I'm interested in any feedback that makes it a better learning tool.

I'll be around to chat!