Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsChungin “Roy” Lee, then a Columbia University computer-science student, built an AI assistant called Interview Coder and said he used it during technical interviews at Amazon, Meta, TikTok and other companies. After he published an Amazon interview demonstration, the offer became disputed, Columbia opened a disciplinary case, and the product later evolved into the startup Cluely.
The episode was not simply a story about a student “beating” Amazon. It combined a hiring-process vulnerability, disputed claims about an employment offer, a university discipline case and the commercialization of covert AI assistance. It also raised a legitimate question for candidates and employers: do coding interviews measure practical engineering ability, or mainly preparation for coding interviews?
As an Amazon Associate I earn from qualifying purchases.
What happened, in brief
Lee said he received an Amazon offer after using Interview Coder during the company’s technical interview process. He later said he rejected the offer because his objective was to demonstrate the product, not take the job. Other reporting described Amazon as rescinding or withdrawing the offer after the company learned about the undisclosed AI use and the public video.
Recommended Free Tools
Amazon did not publicly explain Lee’s individual case. Its stated policy was that candidates had to agree not to use unauthorized tools during interviews. That leaves an important distinction: Lee’s account established what he claimed happened, while the precise internal sequence behind the offer was never fully confirmed publicly.
#1 Best Overall
- Careercup, Easy To Read
- Condition : Good
- Compact for travelling
Shortly afterward, Columbia sent Lee a redacted complaint and scheduled a disciplinary hearing. Later reporting said the university imposed a one-year suspension in March 2025. Available accounts indicate that publication or disclosure of materials connected with the disciplinary proceedings was part of the stated basis, rather than a publicly confirmed finding that the suspension rested solely on building or using Interview Coder in an employer interview. Columbia declined to provide a detailed public explanation, citing student-privacy restrictions.
Gizmodo’s March 4, 2025 report is the main source for the original chronology. Later coverage from TechCrunch, B17 News and Chosun added details about the suspension and the product’s next stage.
Who is Roy Lee?
Lee, whose full name is Chungin “Roy” Lee, was a Columbia undergraduate studying computer science when the controversy broke. He described technical interviews as an inefficient barrier to technology jobs and said he had spent approximately 600 hours preparing for coding interviews.
His criticism was that algorithm-heavy questions can reward memorization, pattern recognition and familiarity with interview conventions without necessarily measuring the work many software engineers perform every day. Real engineering often involves reading an existing codebase, debugging, communicating with teammates, maintaining systems and making trade-offs under changing requirements.
Lee’s broader predictions about AI—including his forecast that much human intelligence work could become obsolete within two years—were his own predictions, not established facts. They are relevant to the story because the product was presented not just as a shortcut through one interview, but as evidence for a much larger argument about AI and professional work.
What Interview Coder did
At a high level, Interview Coder was an AI-assisted application designed to interpret a coding-interview prompt and generate a possible solution or guidance in real time. Lee described a workflow in which the problem was captured, sent to an AI model and returned as an answer that the candidate could use while responding.
Rank #2
The product was also marketed as difficult for interview-monitoring software to detect. That was a product or marketing claim attributed to Lee—not an independently established technical fact. The available reporting does not prove that the system autonomously solved every interview problem or that it reliably defeated every form of monitoring.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallUsing an AI assistant to prepare for an interview is different from using undisclosed assistance during an assessment. Candidates considering any tool should read the employer’s rules literally and ask whether AI, web searches, notes, external applications or other aids are permitted. A tool that violates the stated conditions can invalidate an offer even if the candidate can perform the job in another setting.
The Amazon interview and the disputed offer
Lee recorded and posted an Amazon interview as a product demonstration. He claimed that Interview Coder helped him pass the technical portion and obtain an offer. The recording was later reported as deleted or removed, although copies or excerpts circulated.
Amazon’s public position was general rather than case-specific: candidates were required to agree not to use unauthorized tools during interviews. Amazon did not publicly confirm every detail of Lee’s recording, state exactly how its hiring team handled the offer or adjudicate the competing accounts.
There are therefore two versions of the ending:
- Lee’s version: he completed the process, received an offer and rejected it because the point was to test and publicize Interview Coder.
- Other reports: Amazon withdrew or rescinded the offer after learning about the undisclosed AI use and the public disclosure.
The careful conclusion is that Lee said he received and rejected an offer, while subsequent reporting characterized the offer as rescinded or withdrawn. Saying simply that “Amazon hired him” overstates what the public record establishes.
What about Meta, TikTok and other companies?
Lee also said he used Interview Coder in processes involving Meta, TikTok and possibly Capital One. Reports said that some offers were later withdrawn after he publicly disclosed the tool’s use.
Rank #3
Those claims should remain attributed to Lee unless supported by a direct statement from the companies or documentary evidence. Meta and TikTok did not respond to requests for comment cited in the original reporting. The episode does not independently establish that Interview Coder passed every interview, that it worked equally well across employers or that it alone caused any offer.
Who reported him to Columbia?
Columbia received a complaint shortly after Lee posted the Amazon video. The complaint was redacted, and the public record does not establish whether it came from Amazon, a university-affiliated person, another student or someone else.
The complaint reportedly characterized the conduct as cheating and referred to Amazon’s relationship with Columbia Engineering. That may explain why the headline described someone as having “tattled” to the university, but it does not establish that Amazon itself filed the complaint. The complainant’s identity was not publicly confirmed in the reporting supplied for this article.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why did Columbia suspend him?
Columbia initially scheduled a disciplinary hearing for March 11, 2025, concerning Interview Coder and its possible use in academic settings. EdexLive reported on the early disciplinary chronology, while Lee later described his position in a public thread archived by Xunroll.
Later reports said Columbia imposed a one-year suspension. The available reporting indicates that the stated disciplinary basis involved publication or disclosure of materials associated with the disciplinary process. That is different from saying Columbia publicly confirmed that it suspended Lee solely for using AI in an employer’s interview.
Several questions remain unresolved from public information: the exact jurisdiction Columbia asserted over the conduct, how it distinguished building the tool from using it in an interview or class, what due-process procedures Lee received and which specific acts formed the final basis for discipline. Columbia’s privacy restrictions prevented a complete public account.
Was this cheating?
The answer depends on which question is being asked, and two questions should not be collapsed into one.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →From a hiring company’s perspective, undisclosed real-time assistance during an assessment can misrepresent a candidate’s unaided ability. If candidates agreed not to use unauthorized tools, violating that agreement is a rules issue regardless of whether the interview itself was a good measure of engineering skill.
Lee’s counterargument was that conventional technical interviews are flawed. A candidate may be able to build reliable software yet perform poorly on an artificial algorithm puzzle, while another candidate may have practiced hundreds of similar questions without strong experience maintaining production code.
Both points can be true. Technical interviews may be poorly designed, and an undisclosed assistant may still violate the conditions of the test. Criticizing an assessment does not automatically authorize a candidate to disregard its rules.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the controversy says about technical interviews
LeetCode-style interviews remain popular because they are relatively standardized, scalable and easier to compare than open-ended assessments. They can reveal useful information about data structures, algorithmic reasoning and a candidate’s ability to explain a solution under constraints.
They also have limitations. They may over-reward intensive preparation and memorized patterns while under-testing:
Best Value
- debugging a flawed implementation;
- working inside an unfamiliar repository;
- maintaining and refactoring existing code;
- communicating with teammates;
- making product and engineering trade-offs;
- handling ambiguous requirements; and
- responding to changing constraints.
Potential alternatives include take-home assignments, pair programming, repository-based tasks, debugging exercises and longer project evaluations. None is a perfect solution. Take-home work increases candidate time costs and may be completed with outside assistance. Pair programming costs more interviewer time. Repository exercises raise privacy and access questions. Longer evaluations can create inconsistent grading and favor candidates with more free time.
AI makes the trade-off more urgent. Employers can respond with stricter monitoring, but surveillance can produce false positives, create privacy concerns and still fail to measure practical ability. They can also use follow-up questions, require candidates to defend design choices, test code in a real repository and combine several assessment formats rather than treating one remote coding screen as decisive.
From Interview Coder to Cluely
The controversy became a startup story as well as a disciplinary and hiring story. Interview Coder later became part of Lee’s startup, Cluely. The company broadened the concept into a hidden AI assistant for interviews, sales calls, exams and other interactions, using deliberately provocative language around “cheating.”
TechCrunch reported on April 21, 2025 that Cluely announced a $5.3 million seed round. That figure is startup financing, not consumer revenue or a subscription price, and it does not prove that the product is effective or permitted in any particular setting.
Gizmodo reported Interview Coder at $60 per month in March 2025. That is a historical price, not a verified current price. Prospective buyers should check a vendor’s current terms and, more importantly, determine whether use is allowed by the employer, school or testing provider. A subscription cannot protect a user from an invalidated assessment, a rescinded offer or disciplinary consequences.
What candidates and employers should learn
For candidates
- Ask in writing whether AI tools, search engines, notes and external applications are allowed.
- Use AI freely for preparation only where the rules permit it; do not assume workplace AI policies apply to an interview.
- Do not treat a company’s failure to detect assistance as permission to use it.
- Expect follow-up questions or later work samples to test whether you understand the submitted solution.
- Keep records of the instructions you received if the assessment rules are unclear.
For employers
- State tool restrictions clearly before every stage of the process.
- Keep rules consistent across recruiter screens, online assessments and live interviews.
- Use follow-up questioning to test reasoning rather than relying only on automated monitoring.
- Include practical exercises involving debugging, code maintenance and communication.
- Be transparent about monitoring and avoid treating imperfect proctoring signals as conclusive proof.
For job seekers who want legitimate preparation, LeetCode is relevant for algorithm and data-structure practice. Interviewing.io focuses on mock interviews and live coaching, while Exponent covers areas such as system design, behavioral interviews and product preparation. Their current pricing and features should be checked directly before purchase.
The unresolved lesson
Lee’s stunt exposed a security problem in remote technical assessments: a hidden assistant can undermine what the interview is supposed to measure. But it did not prove that all technical interviews are obsolete, that Interview Coder independently earned every claimed offer or that covert AI use is harmless because engineers may use AI at work.
Free tools Windows power users keep installed
One-click scans. No signup required.
The strongest lesson is narrower and more useful. Employers need assessments that test real engineering judgment and rules that account explicitly for AI. Candidates need to distinguish between preparing with assistance and representing assisted work as unaided performance. A flawed test can deserve criticism while the undisclosed way it was taken still carries professional consequences.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
