Platforms like LeetCode and Codeforces are great for practicing algorithms, but they don't reflect how most software development actually works.
In real projects, developers spend more time reading code, fixing bugs, improving documentation, writing tests and collaborating with maintainers.
At the same time, many developers want to contribute to open source projects on GitHub but often don't know where to start. Large repositories can feel overwhelming and even "good first issues" sometimes lack enough context for beginners.
This talk explores the idea of turning real GitHub issues into structured development challenges. Using the GitHub API together with repository metadata and code structure analysis, a system can summarize repository context, classify issues by difficulty and guide contributors to the most relevant parts of the codebase.
Instead of solving algorithm puzzles, developers would work on real problems from real projects. Contributions could be evaluated using practical signals such as code quality, tests added, documentation improvements and maintainer feedback. Over time, contributors could follow a skill-based path that gradually introduces more complex challenges.
During the talk, I will also discuss how issues from real GitHub repositories can be analyzed and transformed into structured development challenges through a pipeline that includes issue ingestion, repository analysis and challenge generation. An example workflow will illustrate how a GitHub issue can be converted into a guided challenge that highlights relevant files, suggested entry points and task context.
I will walk through the concept, the architecture of such a system and how approaches like this could help developers gain practical experience while making open source contribution easier to start.
Why algorithm competitions differ from real software development workflows
The challenges beginners face when navigating and contributing to large open source repositories
How GitHub issues can be analyzed and transformed into structured development challenges
Techniques for guiding contributors to relevant parts of large codebases
Practical signals for evaluating real-world developer contributions
This sounds useful, but the references don't show me anything really related to the talk.