Cisco hands 27-year-old engineer unlimited AI tokens to refactor 98% of legacy code

What happened

Cisco has handed one of its largest engineering jobs to a 27-year-old and told him to use as much AI compute as it takes. According to Fortune, company president and chief product officer Jeetu Patel has given Padmanabhan Krishnamurthy — "Paddy" — what Patel described as "unlimited tokens" to refactor the company's legacy code, telling him he can "go nuts."

The mandate comes with numbers attached. Fortune reports a four-week window to prove the concept, and a stated goal of refactoring 98% of Cisco's legacy code within 24 months.

An acquired founder stuck five levels down

Krishnamurthy arrived at Cisco in 2025 through the company's acquisition of EzDubs, a real-time translation startup he co-founded in 2023. Despite the founder background, Fortune says he ended up, in Patel's words, "buried away in the organization doing something unstrategic," five levels below Patel in the hierarchy.

The turning point came during a visit by Patel to Cisco's San Francisco office. In conversation, Krishnamurthy argued he could rewrite Cisco's legacy code with a two-person team. Patel told Fortune that even the most AI-forward people on his own staff were not thinking in those terms, and the project grew out of that exchange.

A deliberate bet on early-career engineers

Patel framed the appointment as part of a broader argument about hiring. He told Fortune that if companies stop recruiting early-career workers, they cut off the supply of fresh ideas and lose the questions that only inexperienced people think to ask. He conceded that putting a twenty-something in a room with Cisco veterans of several decades creates friction, but said that is the point: "Tension is good," as long as it stays on ideas rather than personalities.

He described a habit of treating interns as "reverse mentors." One Stanford scholarship student he met at a campus talk, identified only as Baya, was asked to email him monthly with what she was learning about AI — a test of follow-through. She never missed a month, landed a Cisco internship, and redesigned something experienced staff had been unable to crack. Patel told Fortune he did not bring her in for grunt work or data entry, but so she could teach his people to think differently, and he now wants to recruit her.

Patel told Fortune he structures his own team in rough thirds: experienced operators who understand how the Cisco engine works, systems thinkers hired from the market for an outside-in perspective, and AI-native founders, CEOs and engineers who "operate in founder mode."

Others are making the same bet

Fortune's report places Patel alongside other figures who prize inexperience. Thrive Capital founder Josh Kushner told the outlet he specifically prefers hiring people with fewer than four years of industry experience, an approach that produced early stakes in companies including OpenAI. Tastewise CEO Alon Chen said he hires entry-level staff because they bring no preconceived limits to their thinking. And investor Will.i.am, an early backer of Tesla, OpenAI and Pinterest, said his current focus is finding young founders at schools like MIT and Stanford because they are native to AI-era tools.

The backdrop is otherwise rough for the cohort being championed: Fortune reports that roughly 40% of bosses plan to hire fewer graduates on the grounds that AI can do the work more cheaply.

Why it matters

The story is a concrete data point in two overlapping debates.

The first is AI-driven modernization. Legacy refactoring has traditionally been among the slowest, most expensive line items in enterprise IT, scoped in large teams and multi-year budgets. Cisco — a company Fortune describes as a $429 billion giant — is now scoping a rewrite of nearly all its legacy code as a two-person, two-year effort, with AI compute rather than headcount as the main consumable. If the target holds even partially, it resets expectations for what maintenance work should cost and how quickly it can move.

The second is what entry-level engineering work becomes. Patel's stance inverts the common narrative that AI simply eliminates junior roles: at Cisco, the AI-native young engineer is the person entrusted with the highest-leverage project, explicitly assigned to teach veterans rather than absorb low-level tasks.

One caveat is worth keeping in view. The 98%-in-24-months figure is a stated ambition reported by Fortune, not a published result. The four-week checkpoint, and whatever Cisco discloses afterward, will show whether unlimited tokens actually convert into rewritten systems at scale.