Cisco President Gives 27-Year-Old Former CEO Unlimited Tokens, Challenges Him to Rewrite 98% of Legacy Code in 24 Months

Inside a tech giant with a market value of hundreds of billions of dollars and a massive engineering organization, who is qualified to lead a large-scale software rewrite? By conventional logic, the answer is usually an experienced senior engineer, an architect, or a technical leader with decades of enterprise systems development experience.

The 27-year-old former CEO buried inside Cisco suddenly received an impossible mission.

But Cisco President and Chief Product Officer Jeetu Patel handed the opportunity to a young man who was only 27. Even more surprising, this young man proposed that he needed only two people to have a chance at completing an astonishing task: rewriting 98% of Cisco's legacy code within 24 months.

Jeetu Patel not only did not reject the idea, but also gave him a nearly unlimited AI token budget.

"I gave him unlimited tokens because he was going to refactor code," Jeetu Patel revealed in an interview with Fortune. The authorization he gave the young man was quite direct: just go ahead and do it.

The young man is named Padmanabhan Krishnamurthy, and everyone habitually calls him Paddy.

He is not an engineer who just graduated and has no business management experience at all. On the contrary, Paddy once founded an AI startup and served as CEO. It was only after Cisco acquired it that this young founder became mired in the complex organizational system of a large company.

According to Jeetu Patel, at the time he was doing work inside the company that lacked strategic significance. It was not until a chance office conversation that this former entrepreneur regained room to apply his abilities.

On October 30, Fortune reporter Orianna Rosa Royle disclosed this story. In the report, Jeetu Patel not only introduced this bold code refactoring plan, but also put forward a view completely different from current tech industry hiring trends: as more and more companies try to use AI to reduce junior staff, what really needs to change may not be young people's abilities, but the way companies use young talent.

Although the rewrite ratio reaches an exaggerated 98%, this is currently only a goal, not an accomplished result.

According to the Fortune report, Paddy set himself an initial four-week validation deadline, and the final goal of the entire project is to complete the large-scale refactoring within 24 months. Cisco has not yet disclosed the proportion of code completed, the system scope, the technical approach, or test results.

What is truly noteworthy about this experiment is not only whether two people can complete work that traditionally requires a huge team, but also how AI coding is disrupting the way software engineering is organized in large enterprises.

After Paddy joined Cisco, the difference between the two organizational environments quickly became apparent in him.

At the startup, he had been CEO and could directly decide product and engineering direction. After joining Cisco, he did not immediately receive responsibilities matching his previous entrepreneurial experience.

Jeetu Patel described to Fortune that Paddy was buried inside the organization at the time, taking on unimportant work. This problem did not change until Jeetu Patel visited Cisco's San Francisco office.

During their conversation, Paddy proposed a rather radical idea: use AI, with a small team of only two people, to rewrite Cisco's legacy code.

For any large tech company, this is not an easy proposal to accept. Legacy code includes not only software written historically, but may also involve long-accumulated business logic, system dependencies, interface protocols, and compatibility requirements. Behind the code are often large numbers of actual users and running businesses.

Therefore, large-scale refactoring has never been just about replacing old code with new code, especially since the team size Paddy envisioned was only two people.

Jeetu Patel admitted that even the members of the team most enthusiastic about embracing AI had never proposed such a bold idea.

But he decided to give this young man a chance.

Paddy set himself a four-week validation deadline, with the ultimate hope of completing the rewrite of 98% of legacy code within two years. Jeetu Patel provided him with resource support not constrained by conventional token quotas.

Big-company reflection: Why can the most creative young people only do odd jobs?

Jeetu Patel's approach has attracted attention also because it comes at a time when employment pressure on young tech workers is increasing.

At the beginning of the report, Fortune mentioned that a survey of employers showed about 40% of managers plan to reduce hiring of new graduates, one reason being that AI can complete some junior work at lower cost.

Jeetu Patel does not agree with this trend of reducing hiring of young people. In his view, if companies stop hiring employees in the early stages of their careers and let existing teams continue to use the old ways of doing things, where will new ideas come from? And who should challenge entrenched ways of working?

Jeetu Patel does not believe that senior employees lack the ability to innovate; on the contrary, he recognizes the industry experience accumulated over the long term. But he has also found that people lacking experience sometimes raise questions that veteran employees would not think of, because in mature organizations, experience not only means capability, but may also mean a default acceptance of established rules.

For example, when a company is already accustomed to transforming legacy systems with large teams and long cycles, senior engineers tend to first consider system dependencies, risks, and resource coordination, whereas a young person who has just moved from a startup to a big tech company may directly ask: why can't this be handed to two developers and AI?

What Jeetu Patel wants to see is precisely this way of thinking that is not bound by existing processes. He even believes that a certain degree of tension is needed within an organization.

"Tension is good." Jeetu Patel told Fortune.

In his understanding, truly valuable friction should occur between viewpoints and methods, not between people. Only by allowing members from different backgrounds to challenge each other does a team have the opportunity to break through the boundaries of existing work. He wants employees to dare to set seemingly unreasonable goals and to seek solutions beyond traditional methods.

He told Fortune that he treats interns as a kind of "reverse mentor," letting these young people help senior employees engage with new technologies and ways of thinking. He even proactively goes to universities to look for talent.

One recent experience took place at Stanford.

At the time, Jeetu Patel was giving a talk to a group of scholarship students in their late teens. After the event, many students lined up to ask him questions, but a girl named Baya did not join immediately. She stood nearby, carefully listening to the questions others asked, and only began her own questions after everyone else had finished. Jeetu Patel still noticed her.

For Jeetu Patel, one excellent interaction is not enough to fully prove a person's ability. He wants to know whether young people can keep learning and deliver on their commitments. Baya did.

She did not miss a single monthly email, and later obtained an internship opportunity at Cisco.

According to Jeetu Patel, during her internship, Baya redesigned a piece of work that even experienced employees had previously failed to handle well.

In fact, Jeetu Patel's attitude toward hiring interns is very clear: he does not want young people to join the company only to do data entry and trivial work, but rather to bring a different way of thinking to the original team. Now, he already wants to hire Baya formally.

From Paddy to Baya, Jeetu Patel is in fact pushing the same thing: rather than having young people spend several years first adapting to the way a big company works, it is better to have them participate from their first day at the company in changing those ways of working. This differs markedly from the traditional path for cultivating software engineering talent.

In the past, junior engineers often had to start with simple bug fixes, small feature development, testing, or maintenance work, gradually becoming familiar with the business and systems, and then gaining decision-making authority on important projects. But as Coding Agents take over more and more basic programming tasks, the original promotion and training methods also face a new problem: if a large amount of junior work can be done by AI, through what means should young engineers accumulate experience?

So it is also time to rethink the new training path for young engineers in enterprises.

At the end of the interview, Jeetu Patel talked about his approach to managing teams. He does not place young people in opposition to senior employees, but tries to put people from different backgrounds into the same organization.

According to his introduction, the team is roughly divided into three categories.

About one third are experienced management and operations talent familiar with Cisco's internal ways of working; they understand the systems, processes, and business needs of a large enterprise.

Another third are people with an external market perspective who can think systematically; these people can observe industry changes from outside the company and bring different judgments to internal decisions.

Finally, one third consists of entrepreneurial founders, CEOs, and engineers who are familiar with AI technology and can advance work in an entrepreneurial way.

Jeetu Patel believes that when these three types of people act alone, their strength is far less than when they cooperate with one another.

Traditional big companies have mature engineering systems, industry experience, and business resources, but they also easily develop fixed decision-making processes. Entrepreneurs are good at rapid experimentation and redefining problems, but may lack understanding of large, complex systems. AI natives are more natural at mastering new tools, but do not necessarily understand the various risks in the long-term operation of production systems. If these three capabilities can be combined, enterprises may be able to improve the efficiency of technology exploration and delivery while maintaining stability.

This article comes from the WeChat public account "InfoQ" (ID: infoqchina), author: Dongmei, published by 36Kr with authorization.