Big Tech Blacklists Roy Lee For Using The Same AI They Promote

For software engineers, cracking a LeetCode interview is often the key to landing a high-paying tech job. These timed coding challenges, typically lasting 45 minutes, have become widely accepted, if much-debated for candidates hoping to get into companies like Amazon, Meta, and TikTok. Many, including Roy Lee, have questioned the effectiveness of this system, arguing that memorizing coding riddles does little to reflect real-world programming skills.

Unlike most frustrated candidates, however, Lee did more than just complain. He built a tool to exploit the system and one that could instantly solve LeetCode problems, evade detection during interviews, and, in the process, turn the entire interview process on its head. He called it Interview Coder, a $60/month desktop application designed to make LeetCode tests obsolete. The tagline on his website says it all: “F*ck LeetCode.”

_{Credits: Interview Coder}

Lee, a Columbia University student, spent over 600 hours grinding through LeetCode problems to reach the top 1% of ranked users. But the deeper he got into the process, the more it felt like a waste of time. Instead of writing real software, he was memorizing patterns to regurgitate in interviews. The realization led him to automate the entire process.

Interview Coder operates as an undetectable overlay that captures the interview question via a screenshot, processes it, and generates a solution in real time. The window remains translucent and strategically placed so an interviewer can’t see any eye movement that would give it away. The tool also keeps the cursor locked in place to prevent any suspicious behavior. There are multiple small but essential design choices that make it impossible for interviewers to detect.

Lee spent an entire recruiting season refining the tool, testing it against Amazon, Meta, Capital One, and TikTok. He even used it in live interviews, recording his session with Amazon and later posting it online—an act that made his story explode.

The moment Lee shared his Amazon interview, the internet took notice. Within weeks, the story had amassed over 50 million views, propelling Interview Coder’s revenue to $1 million ARR with a 110k MRR, growing at 20% week-over-week. But not everyone was entertained. Amazon, unimpressed by Lee’s stunt, allegedly filed a complaint against him with Columbia University. The school initiated disciplinary proceedings on March 11, putting Lee at risk of expulsion. On X (formerly Twitter), he responded with characteristic defiance, saying:

amazon execs r so mad LOLLL maybe stop asking dumb interview questions and people wouldn't build shit like thishttps://t.co/RMwouy3enr pic.twitter.com/sBdWTx2aFJ

— Roy (@im_roy_lee) February 28, 2025

Lee had tested his tool extensively, and it had worked flawlessly. Thousands of users had already tried it since its launch on February 1st, with many reporting that they had successfully landed jobs using it. Not a single user had been caught.

But Lee wasn’t just getting into trouble—he was also making money. The controversy only fueled more interest in his product. As he put it, “There was definitely a point where I thought I had completely burned my career for 20,000 YouTube views… That feeling lasted about a week—until the story blew up. Once it went viral, that became my protection.”

Recruitment was always a face to face interview but since the pandemic most candidates use AI to ace their answers and solve problems within seconds. Since ChatGPT has become “the thing” developers have been laid off and even leaders believe that AI will be writing 100% of the code in the coming years. 

Hiring managers, predictably, are not thrilled about candidates using AI-powered assistants to ace interviews. Conversations on Reddit have revealed how interviewers have noticed suspiciously perfect answers following long pauses. One hiring manager stated they would rather select a candidate who admitted they didn’t know an answer than one who seemed to magically produce a flawless response out of nowhere.

Anthropic, the AI company behind Claude, updated its job application policies in February to explicitly prohibit candidates from using AI tools during the hiring process:

“While we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process. We want to understand your personal interest in Anthropic without mediation through an AI system, and we also want to evaluate your non-AI-assisted communication skills.”

Amazon has also introduced new measures, requiring candidates to acknowledge that they won’t use unauthorized tools during their interviews. Amazon spokesperson Margaret Callahan told CNBC that the company is taking steps to combat AI-assisted cheating.

But Lee sees this as a failure of companies to adapt. “Everyone programs nowadays with the help of AI,” he said. “It doesn’t make sense to have an interview format that assumes you don’t have the use of AI.”

Lee and his co-founder, Neel Shanmugam, built Interview Coder’s core product in just four days, using Electron, React, Vercel, and the OpenAI API. Most of the early development involved real-world testing during Lee’s own job interviews. Despite running into early DDoS attacks that cost him $2,000 in Vercel hosting and API credits, he managed to keep costs low, spending just $20/month at the start.

His approach to product decisions was simple: tweak features based on what worked best in live interviews. The duo later paywalled the software and added authentication, but the core concept remained unchanged which was automating the LeetCode grind while remaining undetectable.

By the time Interview Coder hit $10,000 MRR, Lee had already blacklisted himself from big tech, faced disciplinary action at Columbia, and was en route to expulsion. But the rapid growth of the tool proved that demand was real.

Now, as Interview Coder continues to grow, Lee is preparing for his next move. He’s relocating from New York to San Francisco, hiring influencers like the “Costco Guys” to market his software, and thinking about what comes next.

BOOM! BOOM! BOOM! BOOM! BOOM! pic.twitter.com/JDsJGs2YzY

— Roy (@im_roy_lee) March 2, 2025

When asked if he worries about engineers losing trust within the industry, Lee didn’t seem concerned. His response?

“If there are better tools, then it’s their fault for not resorting to the better alternative. I don’t feel guilty at all for not catering to a company’s inability to adapt.”