A Columbia University student, Roy Lee, faces a disciplinary hearing after admitting to using an AI program he developed to secure internships at major tech companies like Amazon, Meta, and TikTok. Lee, who plans to leave Columbia regardless of the hearing outcome, argues his program, “Interview Coder,” exposes the obsolescence of the roles these companies offer.
The notoriously grueling interview process for tech giants, often referred to as FAANG (Facebook, Amazon, Apple, Netflix, and Google), involves extensive technical interviews. These interviews typically require prospective software engineers to solve complex coding problems live, often under the watchful eye of a company employee. This process, Lee argues, is more about performance than practical skill.
Lee, a sophomore slated to graduate in 2026, described his 600 hours of interview preparation as a “miserable experience,” ultimately extinguishing his passion for programming. His LeetCode profile, a platform used by programmers to practice for technical interviews, reflects the extensive effort he invested. He criticizes the interview process as an exercise in memorization and performance rather than a true assessment of a programmer’s abilities. He believes the focus on algorithmic problem-solving doesn’t reflect the day-to-day work of a software engineer.
To circumvent this system, Lee created Interview Coder, a program that essentially uses image recognition and ChatGPT to solve technical interview questions. He describes the program as surprisingly simple, claiming it could be replicated in under 1,000 lines of code. The program’s code is publicly available on Github. Lee maintains that his program is undetectable by the monitoring software used by tech companies during interviews.
Lee claims to have successfully used Interview Coder to receive offers from TikTok, Meta, and Amazon. He even recorded his entire interview process with Amazon and posted it on YouTube as a demonstration of the program’s efficacy. Amazon declined to comment on Lee’s specific case, but acknowledged its recruiting process is continually evolving. Meta and TikTok did not respond to requests for comment. Amazon spokesperson Margaret Callahan stated that while the company welcomes candidates to share their experiences with generative AI tools, they must pledge not to use unauthorized tools during the interview process.
After Lee posted the Amazon interview video online, an anonymous complaint was sent to Columbia University, accusing him of cheating. Columbia subsequently scheduled a disciplinary hearing. The complaint, a redacted version of which was forwarded to Lee, alleges that Amazon rescinded its offer due to his actions. The complaint expresses concern over Lee’s alleged cheating and emphasizes Amazon’s long-standing relationship with Columbia University. Columbia University declined to comment, citing federal privacy regulations, and Amazon refused to comment directly on the matter. However, Lee has shared documentation with MaagX.com verifying the existence of the disciplinary proceedings.
Lee, who believes Large Language Models (LLMs) will render many “human intelligence” jobs obsolete within two years, states he won’t attend the hearing. He plans to leave New York City and pursue other opportunities. The publicity surrounding his actions has generated considerable attention within programming communities and boosted interest in Interview Coder, for which Lee now sells subscriptions. He admits his interviews and subsequent posts were partly a marketing strategy, but maintains his conviction that the technical interview process is detrimental to programmers and the broader development ecosystem. He believes it’s an outdated, inefficient system that needs to be reformed.
He argues that big tech companies lack the incentive to change, benefiting from the existing system while negatively impacting the global developer community. He sees the current technical interview process as an obsolete attempt at a standardized test for problem-solving in a rapidly changing technological landscape.
