Aerospace engineer turned builder. I work where technology meets the business, and I'm at my best making the two understand each other.
Aerospace engineer by training, now leading strategy and automation at Brox.AI, an AI consumer research company. I win enterprise clients and build the AI agents that deliver their research. I approach problems through systems and incentives, and I test ideas by arguing the opposing case.
Today: Brox.AI, Strategy & Automation Lead for Turkey & the Middle East. Full experience ↓
Automated quality assurance for client deliverables. Every dashboard is reviewed against the client brief before release, and the client PDF and email are drafted for editor approval.
An agent skill library that automates the research workflow end to end, from client brief and study design through analysis to a branded client report.
Automated recruitment for hard-to-reach research audiences, including commercial banking clients, GLP-1 users, dermatologists and high-net-worth investors. Agents build and launch Meta campaigns via MCP, with full-funnel conversion tracking across 15+ campaigns in four markets.
Consolidates every client deliverable from the CRM into a single timeline per account, and uses the Claude API to recommend the next study to propose.
A systematic review of 62 studies on predictive maintenance with aircraft digital twins, proposing two implementation frameworks. The same principle now underpins my work modelling consumers rather than aircraft.
Conventional threshold monitoring and a digital twin, compared on the same engine. The twin models vibration as a Fourier series and standardises every sensor reading, flagging the first signal to drift. Select a fault and a sensor fidelity to compare detection times.
Group design project: a single aircraft for two assigned routes, Mexico City to Saint Lucia and Rio de Janeiro to Washington. Rio's 1,320 m runway, Mexico City's altitude and demand of 260 passengers a week led to a 140-seat, twin-engine narrow-body.
AI agents · Claude Code & API · MCP · Python · Node.js · Next.js · Apps Script · Google Tag Manager · Pixels & events
Go-to-market · Enterprise sales · Client delivery · Panel growth · Paid social
MATLAB · SolidWorks · Turkish (native) · English (fluent)
Lead client delivery for the NBA, TikTok, Fibabanka and İşbank. Closed $150k in contract value, grew the research panels to 40,000 participants across four markets, and automated delivery end to end.
First hire for a new market. Secured the first client, built the panel and sales pipeline, and managed $30k of paid social that recruited 30,000 panellists.
Assessed around 50 corporate balance sheets to support credit approval decisions.
Credit-bearing, graded. Completing January 2027.
First-class dissertation on digital twins in commercial aviation. Aerospace Engineering student representative.
Delivered five startup events on a £20k budget and supported twenty founders through the programme.
Managed the team budget and business plan, allocating funds across the competing priorities of a thirty-engineer team.
I'm from Turkey, where Western and Eastern civilisations have met for thousands of years. Combining different worlds isn't something we learn there, it's who we are. That's what I bring to work: engineering and commercial aren't two camps to me, they're one way of thinking.
Aerospace training at Manchester: decompose the problem, identify the failure modes, fix them. My research focused on digital twins of aircraft.
I build the AI agents and automation tools the team relies on daily. Any task that repeats becomes a system.
Sold to banks, sports leagues and technology companies, opened a new market from zero, and managed six-figure budgets.
This is where I add the most value: translating a client's problem into something technology can solve, and explaining the technology in the client's terms.
Give me a half-defined problem and I'll turn it into a working plan.
I state my view clearly, and change it when presented with a better argument.
Repeatable work is automated, templated or delegated to a model.
I look for the weak point in a plan before the client finds it.
I own the result, not just the task, whichever way it goes.
Problems larger than the job title, where commercial and technical work are one and the same.
I don't travel to tick places off. I go for the food, the people, the history, the streets, the trees and animals, even the weather. These are the things that stayed with me.
Grew up on the Asian side, a short walk from the sea. A city that's been East and West at once taught me you don't have to pick a side.
The more places I see, the clearer it gets: normal only means where you are and who you're with. I try to leave mine at the gate.
Bali showed me you can get a lot done while living in harmony with the people and the place around you. Work hard, leave things as you found them.
Italian trips, Bordeaux, and trying every seafood I could find in Oban all ended up on my stove. No signature dish, on purpose: every trip changes one thing.
I go looking for the old layers. Standing where the world was centred two thousand years ago makes today's certainties look temporary.
How a street is laid out, which trees grow, what the weather does to a day. A film camera, 36 shots a roll, keeps me looking properly.
Istanbul made me, Manchester taught me to build. Next I want to be around people like me, and shape what's being built around me.
Turkish mountains first, then anywhere with snow.
Still learning to read the water. Same rule as skiing: commit, then adjust.
Same dish, new twist. Sometimes a disaster.
One shot, no delete. Lost on a street in a foreign city, every frame counts: I probably won't pass that way again.
In a new city I head for the streets first. Museums are where the history comes alive: what a place was, before I decide what it is.
I organise events for around 100 people, from sunset gatherings to parties in venues.
I love a good hot take. Sports, politics, AI and tech, philosophy, or a paradox that won't leave you alone. Send me yours and I'll reply to the best ones here.
Two boxes. One holds $10,000. The other holds $1,000,000 or nothing: a supercomputer that is almost never wrong filled it only if it predicted you'd take that box alone. One box, or both?