I Quit My GitHub Job Because AI Breaks Software

Connect and ask me anything: https://www.linkedin.com/in/zen-van-riel/ After 3.5 years at GitHub I resigned today. Not because I hated my job, but because I saw an industry-wide trend that I wanted to do something about. Companies are replacing entire parts of the software development lifecycle with AI agents, including code review, testing, deployment decisions, and architecture. The math is simple: imperfect systems generating enormous amounts of code will produce more bugs than humans can catch. I'm joining an AI research lab to build monitoring systems that keep code quality high as agents take on more of the development cycle. What You'll Learn - Why the "100x developer" promise creates a statistical guarantee of more broken software - What happens when the industry treats pull requests as a bottleneck instead of a safeguard - Why stacking agents on agents ("playing dollhouse") does not converge to perfection - How AI safety roles are compensated and what skills actually matter at the $200k+ level Timestamps 00:23 Why I resigned from GitHub 01:11 SDLC Replaced By Agents 02:40 More AI code = more bugs 03:30 Why agents reviewing agents does not work 04:30 The industry is ignoring the problem 05:25 Rogue AI is not fiction 06:15 Some software can be buggy 07:15 My next career step Why I Made This Video I spent 3.5 years building AI systems at GitHub and saw firsthand how the industry is prioritizing raw code output over software quality. I wanted to explain the statistical reality behind this trend before making a career move to help solve it. #AISafety #GitHub Connect LinkedIn: https://www.linkedin.com/in/zen-van-riel Community: https://www.skool.com/ai-engineer Sponsorships & Business Inquiries: [email protected]