The engineer under the GPUs
Open the DigitalOcean homepage today and the headline isn't about virtual machines anymore. It's about agents, GPUs, inference, and a thing the company now calls the Gradient AI Agentic Cloud. Somewhere beneath that marketing language sits the actual work: schemas, queue jobs, React components, the unglamorous plumbing that turns a slide deck about "AI-native infrastructure" into a button a developer can click. Jahanzaib Khan is one of the engineers writing that plumbing.
His title is Engineer II. It is a modest two-word label for a job that sits at the exact seam where the industry is being rebuilt. DigitalOcean spent a decade winning the affection of indie developers and small teams who found Amazon's console intimidating and its bill terrifying. Now the same company is trying to do for AI agents what it once did for Linux droplets: make the complicated thing feel like a few clean clicks. The people who pull that off aren't the keynote speakers. They're the engineers who know where the bodies are buried in the codebase.
Khan describes himself, without irony, as both an "AI and LLM enthusiast" and a "Ruby on Rails developer." Those two phrases rarely share a sentence. Rails is the framework of the 2010s startup boom - elegant, opinionated, a little out of fashion in a world chasing the newest JavaScript runtime. LLMs are the obsession of the 2020s. Holding both at once is its own quiet statement: the future gets built on top of the present, not instead of it. The agent has to call an API, and somebody has to write the endpoint.
The future gets built on top of the present, not instead of it. The agent still has to call an endpoint, and someone has to write it.
The acquisition that came to Lahore
Most engineering careers are a series of decisions: apply here, accept that, relocate there. Khan's took a different shape. He was working at Cloudways, a managed-hosting company with deep roots in Lahore that built a global business making cloud servers easy to launch and forget. Cloudways had a particular kind of customer - the agency, the freelancer, the small SaaS - who wanted the power of AWS or Google Cloud without the PhD in DevOps. It was a good company doing a useful, unsexy thing.
Then DigitalOcean bought it. The acquisition folded Cloudways and its engineers into a New York Stock Exchange-listed cloud provider, and a Lahore-based developer's employer changed without his desk moving an inch. It is one thing to interview your way into a global tech company. It is another to be carried into one because the work you were already doing turned out to matter to someone with a bigger balance sheet. Khan's path is the second kind - and it says something flattering about the work itself.
There's a tidy lesson buried in that story for anyone outside the usual San Francisco-to-Seattle corridor. The center of gravity in cloud engineering is shifting. The people building the infrastructure the rest of us rent are increasingly not where the conference photos are taken. They're in Lahore, in Lagos, in Lisbon, shipping to the same git repositories as everyone else and getting acquired into the same cap tables.
What "agentic cloud" actually means
Strip away the buzzwords and an agentic cloud is a simple, ambitious idea: instead of renting a server and configuring it yourself, you describe what you want an AI agent to do, and the platform handles the GPUs, the model, the memory, and the glue. DigitalOcean's bet is that the next wave of builders won't want to assemble that stack from a dozen vendors. They'll want one place that already speaks the language of agents.
Khan has spoken publicly about building exactly this - the experience of constructing an agentic cloud at DigitalOcean, the kind of talk where an engineer translates months of unglamorous decisions into something an audience can follow. It's worth noticing what that signals. Companies don't put their internal infrastructure on stage unless they think the people who built it can be trusted to explain it well. The engineer who can both write the system and describe it is rarer than either skill alone.
An agentic cloud is a promise: describe the work, and the infrastructure disappears. Someone has to make the infrastructure disappear.
A full stack and a curious mind
Read Khan's tools as a sentence and you get a fair portrait of the man. Ruby on Rails for the backbone. React.js and Next.js for the interface. Heroku and AWS for the deploys. It's the resume of a generalist - the engineer who can carry a feature from database column to rendered pixel without filing a ticket for someone else to finish it. In an era of hyper-specialization, the full-stack builder remains the person a team can point at a vague problem and trust to return with something that works.
What's notable is the curiosity layered on top. Plenty of engineers treat AI as someone else's department - a model team, a research org, a black box upstream. Khan put "LLM enthusiast" on his own description, which is a small act of claiming the territory. It suggests an engineer who, when the industry pivoted, didn't wait to be reassigned. He leaned in. The Rails developer learned to speak fluent agent.
His education started at Gomal University, where he studied computer science - a long way, geographically and otherwise, from the cloud-engineering postcards of the Bay Area. Before the DigitalOcean chapter and the Cloudways chapter, there was earlier work at a Lahore software shop, the kind of place where young developers learn to ship by shipping. None of it reads like a fast track. All of it reads like someone who got good by doing the work, repeatedly, in public, until the work brought the bigger opportunity to him.
The quiet kind of important
There's a particular type of engineer who never trends and never needs to. They don't ship a viral demo. They ship the thing under the demo - the part that has to not break when ten thousand people show up. Khan looks like that type. The agentic cloud will get the headlines. The launch announcements will name the product, not the person who wired the endpoint that makes the announcement true.
That's the deal these engineers accept, and there's a dignity in it. The reward isn't applause. It's the quiet knowledge that something real exists in the world that didn't exist last quarter, and you are the reason it works. Multiply that across a career and you get the actual infrastructure of the internet - built by people most users will never name.
So here is Jahanzaib Khan, mid-stride: a Lahore engineer who learned Rails when Rails was king, picked up LLMs when the ground shifted, got acquired into a global cloud because his work was worth acquiring, and now spends his days making the most hyped technology of the decade boring enough to be useful. Not a bad place to be standing while the rest of the industry argues about what comes next. He's already busy building it.