"All the users we have are not from the tech world — so we're not relying on them to tell us what to build."

Ask Shahin Katebi what he is, before what he does, and the answer is disarmingly simple. "I see myself as a builder. I really like to create things, make things from scratch." It is the thread that runs backward through almost twenty years and half a dozen countries. He started as a software engineer in 2005, went freelance, and by 2009 had founded his own software agency in Iran — one of the first shops in the region doing mobile development in the earliest days of iOS and Android. There weren't enough engineers to hire, so he made more of them: training people, running workshops, organizing conferences on mobile and UX at a time when all of it was new. When he wanted a startup scene, he helped build that too, running Startup Weekend events for years.

From there, the map naturally widened. Scaling software and connected electronics at his first agency in Iran led to building US ventures in logistics and healthcare in 2018, followed by two years in Estonia at cybersecurity firm Malwarebytes. Moving to London, he spent two years at Meta building Instagram's first blockchain and NFT products. By 2023, he was back in the startup ecosystem — architecting applied-AI stacks and co-founding VitalStaf, restless in the productive way that founders are.

That restlessness isn't only professional. Away from software, Shahin designs electronics and 3D-printed objects, and he has a particular, almost contrarian taste: smart devices with no screens. "I don't like these devices to have LCDs or displays or LEDs," he says. "I want more natural, nature-friendly representation — like a stone phone would look." It is a small tell about the whole man — a builder who keeps asking whether the obvious way of making a thing is really the right one.

"I see myself as a builder. I really like to create things, make things from scratch."

Today that instinct is spread across several brands, because Shahin doesn't believe in a single one that contains him. "I usually brand myself as a tech entrepreneur — there's no single brand that represents me. It would be like me representing different brands." The studio is Yeka Studio. Separately, he co-founded VitalStaf — an independent startup attacking a stubbornly analog problem: nurse staffing in the US, where recruitment still runs "manually by phone, by spreadsheets." VitalStaf pulls clinicians, agencies and healthcare facilities into one place for the entire lifecycle — applications, compliance checks, offers, timesheets, payroll — plus an AI recruiter that matches travel nurses to their next thirteen-week posting. Newer still is Prodit, and newest of all is London Venture Kitchen, an embedded execution partner built to deliver the hands-on leadership and delivery support founders keep asking him for.

If there is a single idea underneath all of it, it is where a human belongs in the loop — and Shahin has strong, specific opinions about that, because he's earned them. Prodit is his AI-assisted development workflow: agents translate incoming business requirements into technical terms, open pull requests, groom Jira tickets, ship. But the design isn't "let the AI run." It's knowing precisely where an experienced operator has to stop and check. "It's an AI-assistant workflow for software development, with real humans in the loop — I know exactly where we should put those humans in the loop." Fifteen years of running agencies is what tells him where. And rather than bolt a new interface onto his team, the agents work through the tools people already trust — Jira, Confluence (feedback left as emoji reactions), GitHub, n8n. "Basically everything is done just like how a human would use the platform. I didn't want to introduce a new UI to our team."

"I know exactly where we should put those humans in the loop."

Only at the end does the conversation reach the number — and here Shahin is at his most distinctive. VitalStaf's users are travel nurses and staffing agencies, not product managers; they don't arrive with feature requests, so he doesn't pretend they do. "All the users we have here are not from the tech world, so we're not really relying on getting actual feedback from them." What decides the roadmap instead is domain knowledge — a co-founder with fifteen years in healthcare recruitment. Product data isn't ignored; it's given a narrow, trustworthy job. Microsoft Clarity and Mixpanel session recordings show him exactly where a clinician stalled in an early user cohort, and that isn't a statistic to optimize — it's a reason to make a call. "When we see someone in onboarding stopped somewhere, we contact them and try to understand what the problem was." The data points; the human closes the loop. It is the same conviction as the nail he keeps hitting everywhere else — that speed is only worth having if judgment is still standing where it matters.