Launching today
Swenest
A platform that teaches how to navigate real-world code.
25 followers
A platform that teaches how to navigate real-world code.
25 followers
Swenest stands out with real merged GitHub PRs as tasks, not artificial problems, symptom-only descriptions; users find the cause themselves, progressive AI hints that guide without spoiling, AI judge grading against the actual merged fix and XP-gated difficulty progression. Most platforms don't train navigating unfamiliar, real-world codebases with structured guidance with a gamified platform.
Swenest
GPT 5.6 models (Terra and Luna) in Codex were used for building and testing the application end-to-end; from prototyping, to ensuring it works in production. Codex also helped with automation of the tasks data seeding/curation for the platform. It leveraged the browser tool for automation testing as well.
Swenest
Lot of aspiring Software engineers, juniors and students find it difficult to work on or contribute to open source projects for getting real world experience. It takes a lot of time and effort to find a repository with unassigned issues that they can use their foundations to work on. Also in this AI era, it's important skill to be able to read the code that's not written by you. There aren't many platforms that helps them get that confidence and teaches them how to navigate real world codebases with hands-on learning experience. So, I built Swenest, a gamified learning platform for aspiring/junior SWE's and computer science students that helps them fill this gap.
This is a strong idea. Real merged PRs are much closer to how engineering actually feels than isolated toy problems. The symptom-only setup plus hints also sounds useful because half the skill is figuring out where the bug even lives.
I’d be especially interested in how the AI judge handles valid alternative fixes, since real code usually has more than one reasonable path.
Swenest
Hi @vahid_davoudi, thanks for the feedback! Currently, the LLM judge gets all the context about the problem; i.e. it gets full context about the merged PR (the fix, diff, and reasoning), so it evaluates intent rather than just exact matches. Handling genuinely alternative fixes is still a weak spot though, especially with the light weight model used for the current demo.
The plan is to add sandbox execution so the judge can actually run and test submitted code rather than just reason about it; that's next on the roadmap.
Glad that the idea resonates, that's the core insight Swenest is built around :)