Giles Strong spent his early career helping optimise the vast machines that measure the universe. As a Senior AI Research Engineer at Tektome, he is now pointing the same instinct at a more down-to-earth problem: how buildings get designed.
A physicist who grew up wanting the universe.
Giles knew early what he wanted to be. Through school and into university, the goal was to become a quantum physicist, and he visited CERN as a child more than once, long before he would work there. He studied theoretical physics, then took a postgraduate degree in experimental particle physics, and from there joined the ATLAS collaboration at CERN. His role there leaned less towards analysing collision data and more towards the computing behind it.
Teaching particle physics to think with deep learning.
He moved to Lisbon for a doctorate that set out to bring deep learning into particle physics, this time within the CMS collaboration, just as the field was really beginning to experiment with neural networks. His objective was simple to state and hard to do: work out how deep learning could be incorporated into the discipline. The programme also placed him in a Europe-wide network of ten doctoral researchers, each paired with an industry partner. His placement sent him to Moscow, where he spent three sub-zero months helping build a platform for citizen-science data competitions.
After the PhD, the network’s coordinator invited him to Padua, home to one of the oldest universities in the world, for a postdoc. The detectors that study particle collisions take decades to design and build, and rely heavily on heuristics to get right. Giles wanted to change that. The aim, in his words, was an end-to-end simulation of an experiment: the physics, the processing, and the statistical analysis that follows, all modelled together, so a team could ask whether a design change would improve sensitivity and how long an experiment would need to run to yield new discoveries. In practice, it meant building software to optimise detectors for specific uses.
From detectors to designs to buildings.
That is the thread worth noticing. The problem never really changed, only its subject. Giles had been posting his research publicly when a startup, Braid, reached out to say they liked it. He became one of their first researchers, applying the same optimisation methods to industrial design.
While there, he began looking at where to go next. He found a shortlist of companies, Incubit and Tektome among them, and emailed his CV directly. Naoki replied quickly, a colleague followed soon after, and Giles joined Tektome as an AI Research Engineer. The move brought far more software engineering than research had asked of him. Writing code for clients, where reliability and scalability are the whole point, is a different craft from writing code for a paper, and he closed the gap the way he tends to: by building things himself.
What he found here was a way of working that suited him. Engineering at Tektome runs on a rhythm the team calls Forge: three weeks of focused, heads-down building, then a lighter week of cooldown to rest, learn, and prototype the next idea. Anyone can pitch a piece of work, with no seniority gate, and whoever takes it on owns the whole loop, from talking to the user through design, build, testing and deployment. There is no separate project manager, designer, or QA sitting between the engineer and the person they are building for. Tektome has a phrase for it: everyone is a product engineer.
How engineering works here
- Pitch any idea as a bet, with no seniority gate
- Own the whole loop, from the user through to deploy, with no handoffs
- Three focused weeks of building, then a week of cooldown to rest and explore
It is a way of working he clearly enjoys: proposing ideas, owning them end to end, and being trusted to run with them rather than waiting to be handed a task.
A few questions for Giles.
We put a couple of those directly to him: why AEC, and where he thinks all of this is heading.
What are you working on at Tektome, and what drew you to the AEC industry?
There’s a lot of digital-transformation happening in Japan right now, with a big push from government too, and AEC is one of the major industries, so it’s natural that it would want to become more digital and bring in AI. The framing matters to me, though. It’s not like you’re trying to replace anything, you’re thinking about how it can be best offered to the industries. It’s about what problems they face, and it’s more focused on coming up with original tools and how those can help them. What we’re building is a very flexible platform that allows them to build their own dedicated apps to best address their specific, individual problems.
What excites you most about where this work is heading?
The adoption of AI feeds on itself. As more people take it up, and use it to build AI products of their own, the whole thing becomes a cycle that keeps accelerating. You can already see it happening: even the companies building the large language models are using their own models to develop the next generation.
Now we’re in a situation where you can give a vague goal of what you want to create, and the AI will go back and create something for you as a starting point.
Giles Strong, Senior AI Research Engineer, Tektome
It is a fitting place for someone who has spent his career optimising complex systems to land: a field where the tools are starting to help build themselves, pointed at an industry that builds the world around us.
Join the team
We’re a team where research-grade thinking meets real production work, and where you own what you build, end to end. If that’s the kind of place you’ve been looking for, take a look at where you might fit.