Hercy Chang and the Art of Showing Your Work

Hercy Chang had a university major to negotiate. His father preferred business. His mother preferred art. He wanted computer science. The family arrived at an unusually practical compromise: if he could make money from programming, he could pursue it. His father supplied $99 for an Apple developer registration. In the first month, Chang earned back the fee and several hundred dollars more. An argument about the future had acquired a receipt.

It was their second wager. Earlier, admission to a key high school had earned him a MacBook and an iPhone. He used the computer to learn software development. These were small, domestic transactions, years before the corporate language of talent assessment entered the picture. Yet they make a useful opening to his story. Asked to justify his interest, he produced something that worked. The demonstration settled the question.

Chang, known in Chinese as Zhang Yunhao, went on to study computer science at the University of Arizona. He had begun developing software at 13, working across web technology, cloud computing and mobile applications. The sequence matters: making things preceded the qualification. A degree supplied a formal description of an interest that was already being exercised. His later business would ask employers to look more closely at that distinction.

The connection is an editorial reading of his career, rather than a founding prophecy. A teenager earning app income does not automatically become an entrepreneur in technical education. But there is a recognizable thread between showing parents a useful result and asking companies to judge candidates through actual work. In both situations, the person making the decision needs evidence. A polished explanation can only carry the applicant so far.

The applicant joins the practice room

In 2017, after university, Chang joined LeetCode and worked on improving its product. Winston Tang had established the platform before Chang arrived; Chang became its co-founder and CEO. That distinction gives the story its shape. He was joining an existing project with an audience and a purpose, then helping turn it into a more developed service. By the end of that year, its worldwide user count had passed one million.

The attraction of the service is easy to understand without pretending that a coding interview resembles an ordinary afternoon at work. A candidate faces a bounded problem and has to make progress under scrutiny. Practice offers an earlier encounter with that kind of demand. It gives the learner somewhere to make mistakes before a hiring decision depends on the answer. Even a modest improvement in familiarity can change the experience of approaching a blank editor.

Chang's public speaking placed that practice in a longer career. At QCon Beijing in 2020, he was listed to discuss engineers' growth in an age of algorithms. His speaker biography described a developer concerned with technical education and the future use of technical talent. The ambition extended beyond getting somebody through a single appointment. An interview might create the deadline, but the underlying knowledge had uses after the calendar invitation disappeared.

That creates an interesting demand on the operator. The candidate wants help with an immediate hurdle. The educator wants learning that lasts. The employer wants evidence that the candidate can solve problems. One platform can serve all three, but their wishes do not become identical simply because they share a browser window. Chang's subsequent decisions make more sense when those different customers remain visible.

A Chinese name, a different set of demands

In 2018, Chang established LeetCode's Chinese operation under the name 力扣. Localization brought changes to the product and a team to build them. By May 2020, the China team had grown from three people to roughly 50; about 80 percent worked in product and technical development. The operation had also reached profitability within its first two years. These are milestones from that period, rather than a description of its size today.

The name changed, and so did the scope of the job. A practice platform in another language still has to contend with the institutions around its users. How do companies evaluate an applicant? Which part of hiring happens online? What does a customer expect one supplier to handle? Those questions can determine a product as thoroughly as the wording of a problem. Translation is a beginning; a working service needs more local decisions.

Chang argued for bringing a skills assessment earlier in recruitment. His objection was concrete: a résumé and a conversation about technical knowledge can leave practical ability obscure. He compared the idea to looking at a designer's portfolio before making a judgment. “Programming is a craft,” he said in 2020. The short formulation, translated from Chinese, puts the act of doing something at the center of the evaluation.

“Programming is a craft.”

Hercy Chang · 2020 · translated from Chinese

His argument also involved an admission about the limits of an interview. A few rounds cannot easily produce a complete judgment of a person. The sensible reading of his approach is therefore to make the available evidence more useful. A working answer, an explanation and a discussion of choices give the interviewer something to examine. The person on the other side of the table still has to exercise judgment.

Now the interviewer is a customer

On April 29, 2020, LeetCode announced an expansion into enterprise services in China. The offering addressed technical assessment, recruitment and the development of employees' skills. It brought the company's tools into the employer's process, extending the business beyond individual preparation. The operator of a practice room was now selling equipment to the people conducting the examination.

There is an appealing symmetry to that arrangement. Learners need a way to practice and demonstrate ability. Employers need a way to inspect it. Shared tools can make the exchange easier to organize. They also put considerable responsibility on the people building the tools: a convenient measure is liable to become an influential one. The commercial opportunity and the question of what counts as ability travel together.

By 2021, Chang was describing a growth model in which individual users helped introduce the product to businesses. He compared that approach to Figma. In December of that year, he said LeetCode had more than four million programmer users in China. The comparison concerned how a product could spread into an organization through people who already used it, rather than relying entirely on a sales conversation at the top.

In that same discussion, his ambitions reached beyond conventional software job titles. Product managers, operations staff and people in quantitative finance were among the groups he expected to need programming skills. That was a forecast about the market, not a tally of people already served. It helps explain why he talked about common standards: a broader population of learners would also mean a broader population of people trying to judge their work.

He had a comparatively restrained description of other businesses in the field. In 2021, he treated them as participants working toward more efficient recruitment. His own competitive argument emphasized the community around LeetCode: users practicing, sharing and producing technical content. An answer library can be copied more readily than the habit of returning to one. For an operator, that distinction makes the surrounding community part of the product.

The founder goes back to class

Selling to employers required skills beyond building for individual programmers. Chang joined the third cohort of Tencent's SaaS accelerator, selected from a field of 1,100 applicants for 45 places. His reasons were straightforward: learn from others and find potential collaborators. LeetCode had experience with individual users; acquiring business customers required another kind of knowledge. He was willing to describe that gap plainly.

Tencent's own hiring needs and its network of partner companies offered possible points of connection. The accelerator also put Chang among founders dealing with similar business questions. The scene is a useful corrective to the effortless version of startup expansion. An audience of programmers does not automatically supply a method for selling enterprise software. Someone still has to learn how the purchase gets made and which relationships make the service useful.

In December 2021, LeetCode announced nearly $10 million in Series A financing from Lightspeed China Partners. The planned uses included product improvements, ecosystem development and exploration of markets beyond China and the United States. At the time, the platform reported more than 100 million code submissions a year and hundreds of paying corporate customers. The figures describe the business at that financing moment; they are not current totals.

Submissions are an instructive unit for this particular company. They count attempts, rather than simply registrations. An attempt can fail, teach something and lead to another attempt. A hiring decision ultimately needs a conclusion; a learning tool needs to keep the process open long enough for improvement. Running a business across those two activities means managing different expectations about when an answer should be final.

Recognition, and the next unfinished problem

In 2022, Chang appeared on Cyzone's list of 30 entrepreneurs under 30. The selection considered management, innovation, growth potential and industry influence. That recognition placed a founder working on technical careers alongside entrepreneurs building businesses in other fields. It also offered a rather different form of assessment from the timed programming exercises associated with his company: a judgment of an operator and an organization over time.

Hurun also included Zhang Yunhao in its 2022 China Under30s recognition and featured his work on programmer learning and employment. Its account described a service connecting technical study, competitions, interviews and employer referrals. The range is revealing. The business surrounding a coding exercise had expanded to include several stages of a professional life, each with its own buyer, user and reason to return.

Chang's public LinkedIn activity provides smaller glimpses of that operating work. He has promoted the ability to code in the LeetCode app, credited the app team and thanked users. He has also shared the company's interviewing tools. These are product announcements rather than grand declarations, and that scale suits the subject. Somebody has to make the next feature usable before an ambition about technical careers becomes useful to a candidate.

His stated aim has been to support programmers' continuing professional growth. Read alongside the early family wagers, the ambition has a pleasing continuity. A young developer once needed a way to show that his interest could produce a result. The CEO's business supplies other people with repeated opportunities to demonstrate something, receive feedback and improve it. The audience is larger now. The obligation to show the work remains.