Winston Tang / Founder of LeetCode From failed interviews to a global practice room Malaysia → United States → Amazon → Google → LeetCode Winston Tang / Founder of LeetCode From failed interviews to a global practice room Malaysia → United States → Amazon → Google → LeetCode

Founder profile / Developer education

Winston Tang Turned the Interview He Couldn’t Pass Into a Platform Millions Practice On

A string of failed technical interviews became the blueprint for LeetCode. Winston Tang built a global practice room for software engineers, then watched AI reopen the question at the heart of his career: what, exactly, makes an engineer valuable?

The founding story of LeetCode begins with an error message delivered by a human being. Winston Tang wanted to become a software engineer. The companies he interviewed with kept returning the professional equivalent of “wrong answer.” He had moved from Malaysia to the United States in 2005, carrying ambition but not the network that often tells a candidate what the exam is really testing. His preparation was uneven. The interviews were not kind. Multiple attempts at top technology companies ended in rejection.

There are glamorous founder myths involving garages, lightning bolts and suspiciously photogenic whiteboards. Tang’s origin story is more useful. He had encountered a system whose rules were visible only in fragments. Technical interviews demanded a particular fluency in algorithms and data structures, plus the composure to produce it on command. A computer-science education helped, but the candidate still needed rehearsal. Tang understood the gap because he was standing in it.

So he began building the practice room he wished he had entered before the interview room. LeetCode took a stressful, private ritual and made it repeatable: read a problem, choose a language, write a solution, submit it, inspect the result and try again. The premise was almost offensively plain. Its consequences were not. What had looked like mysterious hiring folklore could become a daily discipline.

01 / The rejection becomes a specification

The problem hiding inside the problem

Tang did eventually win the jobs he had chased, working as a software engineer at Amazon and Google. Yet he continued expanding LeetCode. That choice explains more about him than a polished list of employers. Getting through the gate did not make him forget how opaque the gate had been. He kept working on the map.

“I wanted to create a platform that would help others navigate this process more effectively, and that’s how LeetCode was born.”Winston Tang

LeetCode grew beyond a question bank. Solutions made different approaches legible. Discussions let strangers compare reasoning. Mock interviews added pressure without consequence. Weekly and biweekly contests turned practice into an appointment. Company-oriented problem sets gave candidates a sharper sense of the terrain ahead. The site became part classroom, part gym and part very polite gladiatorial arena. Instead of lions, there were edge cases.

A difficulty, turned into infrastructure

2005Moves from Malaysia to the United States
Early yearsRepeated technical-interview setbacks
2011Begins the platform that becomes LeetCode
2015LeetCode established in Silicon Valley
2021Roughly $10M Series A

The community mechanics mattered. In one public post, Tang asked users to critique a redesigned interface and vote on one another’s feedback so his team could see what to fix first. In another, he welcomed competitors to a weekly contest with no great speech, just an invitation to share the experience. He also posted a logic puzzle about boxes of gold and two magic wands that explode upon touching an empty box. Pedagogy, in his hands, leaves room for a little peril.

Tang has named resilience, vision and “user first” as the traits that guided the company. Such words are common enough to be printed on office walls and ignored beneath them. His actions give them a less ceremonial meaning. Resilience meant improving the platform when growth was slow or the technology misbehaved. Vision meant imagining developers learning together across borders. User first meant remembering the candidate who did not know what everyone else appeared to know.

Winston Tang standing in a sunlit field
Winston Tang built LeetCode from an awkwardly precise memory: arriving at a technical interview without enough of the map. Photograph: Paul Clarke.
02 / Practice crosses borders

A global room with a submit button

LeetCode was formally established in Silicon Valley in 2015 and entered China in 2018. By the time the company announced its Series A financing in 2021, it said its Chinese service had more than four million users, supported 14 programming languages and was used by more than 100 companies to identify technical talent. The roughly $10 million round, led by Lightspeed China Partners, was earmarked for core products and international expansion.

2015Silicon Valley company established
4M+Users reported in China in 2021
$10MApproximate Series A announced in 2021

The scale is striking, but the more distinctive achievement is cultural. “LeetCode” became a verb among job seekers. People speak of grinding it, streaking through it, resenting it and returning to it. A platform intended to reduce uncertainty developed rituals of its own. There are rankings, badges and the small private drama of a solution that passes 96 test cases before failing the 97th. Tang had made the hidden curriculum visible. Inevitably, people began arguing about the curriculum.

He does not claim that solving a neatly framed algorithm problem is the whole of software engineering. Real work is littered with incomplete requests, legacy decisions, social trade-offs and meetings in which the requirements change before the coffee cools. LeetCode concentrates on the foundations beneath that work. Tang compares those foundations to the principles of physics and materials science that an architect must understand. A tool can help with the drawing. It cannot decide whether the building should stand there.

The LeetCode learning loop

Meet a constrained problem
Build and submit a solution
Read the result and other approaches
Return with a better mental model

This focus also explains the business. Much of LeetCode is free. Premium subscriptions provide more comprehensive material and features. Tang has acknowledged the arrangement with refreshing economy: he has “some skin in this game,” and more users are merrier. The aside is funny because it refuses the pose of disinterest. He believes in broader access to technical education and also runs the company that sells a deeper version of it. Both statements can compile.

03 / The machine learns to code

When the answer generator enters the interview

Then generative AI arrived, producing respectable snippets in seconds and reopening an uncomfortable question: if a machine can write code, why keep practicing how humans solve coding problems? For the founder of a coding-practice company, this is not an abstract debate. It reaches directly into the product, the profession and the subscription page.

Tang’s answer is that coding and software engineering are not synonyms. AI is strong at repetitive work, rapid processing and structured tasks. Engineering also requires a person to interpret vague needs, understand a business context, collaborate across specialties and notice the ethical or social consequence that a clean technical answer may ignore. The job contains code, but it also contains judgment about the world into which code is released.

“Software engineers are akin to authors, using tools to articulate their vision while relying on their intrinsic creativity and expertise.”Winston Tang

His view is not that AI changes nothing. He expects some tasks to disappear and pressure on certain entry-level roles to increase. He also expects greater demand for specialists in machine learning, data and computer vision, and for engineers able to handle complex, poorly defined challenges. This is a founder’s optimism, but not a complacent one. The ladder is moving. Tang’s instruction is to keep climbing.

LeetCode has brought AI inside its own operations. Machine learning and natural-language processing support personalized problem recommendations. AI tools help with first-tier assistance and analysis, offering quick responses across different scenarios. Tang says these systems improved efficiency, lowered operating costs and helped retention. The company is teaching people how to remain useful alongside AI while using AI to teach them more precisely. There is a pleasing recursion to it.

He pairs the enthusiasm with familiar cautions: algorithmic bias, user privacy, fairness and the prospect of job displacement. His preferred response is not to freeze the technology but to invest in training, ethical practice and continuous adaptation. The important skills, in his account, mix technical understanding with creativity, problem-solving and ethical reasoning. Languages and frameworks age quickly. A sound mental model has a longer warranty.

04 / Another attempt

The founder who kept the wrong answers

Tang’s favorite life-lesson quotation is the line attributed to Albert Einstein about opportunity sitting in the middle of difficulty. It is perilously close to motivational-poster territory. Tang rescues it by having built the poster into a working product. The difficulty was interview preparation. The opportunity now has test runners, discussion boards, contests, assessments and a button labeled Submit.

When asked what movement he would start, he chose lifelong learning and continuous self-improvement. Again, the language risks floating away. LeetCode pins it down. A practice problem has edges. It can expose a gap, reward a breakthrough and make yesterday’s confusion available for inspection. Continuous learning is no longer an annual resolution. It is Tuesday evening and the solution has exceeded the time limit.

The sharpest part of Tang’s story is not triumph over rejection. Rejection does not automatically contain wisdom; often it contains only rejection. His gift was attention. He remembered exactly what made the process hard: missing connections, inadequate rehearsal, a standard that seemed obvious only after someone had already learned it. He turned those details into product requirements.

Now the profession faces a fresh version of the same anxiety. Engineers are again wondering which skills count, what the evaluator knows and whether the next gate will open. Tang’s answer is consistent with the platform he built: strengthen the fundamentals, use the new tools, study the feedback and continue. The machine may suggest the next line. Someone still has to understand the problem.