This is the eighth in a series of posts about how I ended up where I am today.
The learning system I’d built caught the attention of the Watson team. That, combined with not wanting to spend the rest of my career supporting an email product, got me a role in the Watson group as delivery. My job was to determine what customers needed and deliver an AI solution to meet it.
Watson in the early days felt like a startup inside IBM. The technology was changing day to day. Approaches would pivot without warning. The UIs for figuring out why the AI was misbehaving were hostile at best. Nothing was settled, nothing was polished, and the whole team was high-performing in a way that made you want to keep up. There was something about the energy of it that made me want to know more.
The learning curve was brutal. Months of it. It got to the point where I genuinely thought it would never end. Every time I felt like I had a handle on something, the ground shifted again. I raised this with my manager.
His response was, “Enjoy this time while you can.”
He wasn’t wrong.
Every Couple of Days, a Different Country
Once the projects started coming in, they came from everywhere. Different geographies, different industries, all at once. It was intense. Every couple of days I was in a different country. Europe, Australia, the Middle East. You’d land, get context on the customer’s problem, figure out what Watson could actually do for them, and work with the local teams to deliver it. The local teams were brilliant in their own right. My job was to help them become self-sufficient so the solution didn’t leave when I did.
The pace was relentless but I loved it. Every project taught me something new about how AI met the real world. The gap between what the technology could do in a demo and what it could do in a customer’s environment was where all the interesting problems lived.
Dubai
My second-to-last project on that team changed everything. I was assigned as Technical Lead to build an AI chatbot for the Department of Economic Development in Dubai. Its purpose was to help people start a business. Walk them through the process, answer their questions, point them to the right services.
The technology was still in its infancy. I was given a team that had limited to no knowledge of AI, and newer still were the customer’s team who would eventually take over and run it.
We made it a success. But the part I’m proudest of is what happened to the people. Everyone who worked on my team grew their careers from it. They learned valuable skills, built networks, formed relationships with people that mattered in and out of IBM. Some of them went on to do things I couldn’t have predicted when we started.
The project impressed the government and the local department enough that I got asked to help lead an AI Lab in Dubai. A joint venture with the Dubai government to accelerate the use of AI across different departments.
That was not a role I could say no to.
Building an AI City
I moved to Dubai. A small footnote that turned out to matter: if I hadn’t sat the City and Guilds in C and Unix all those years ago at the printing company, I wouldn’t have been allowed to work in the country. Dubai is strict on having university-level or equivalent qualifications. That two-year course I took out of frustration ended up being the piece of paper that got me through the door.
The education question followed me there too. In one meeting, a presenter asked everyone with a PhD to raise their hand, then said only those people would understand what came next. What followed was a walkthrough of machine learning principles I already knew. There was another “why are you here?” moment when I explained my background. But as before, the work spoke for itself.
I don’t have a hatred for certifications or titles. I’ve helped people get their PhDs. I’ve read people’s published papers to understand what they’re capable of. I judge people on their merit, the same way I’d want them to judge me. But someone hiding behind a title, using it as a wall instead of just explaining their viewpoint and evidence, that does more damage than they realise.
The AI Lab’s mission was to work with government departments to identify where AI could make a difference, score those use cases to see if the technology could actually deliver, and then build the ones that had legs.
We met with departments across the government. Over 300 use cases came out of those conversations, spanning more than ten departments. Around thirty of those were built out as proof of concepts. A handful made it to production. On top of the build work, we were enabling government teams in AI technologies so they could carry things forward themselves.
It was an incredible time. Building an AI city wasn’t just a government directive. It was something everyone was working towards. I got to work with brilliant people and technology I’d never touched before. Robots, drones, IoT, things that made the conversational AI work feel like one small piece of something much larger.
During my time there I helped three people work towards their PhDs and helped another build a package to prove they qualified as an experienced data scientist. That last one carries some irony. The same qualification I helped them put together was one I couldn’t get myself, because my education level didn’t meet the entry requirements.
We’d started expanding to Abu Dhabi and across the UAE when Covid hit and changed the world. I moved back to Ireland and into a new role.
One thing I noticed before I left. In 2017, some of the use cases we’d scoped required the most powerful technology and models available just to get close to working. By 2021, many of those same use cases could be done by a student on a laptop. Technology was accelerating in a way that was hard to fully appreciate while you were inside it.
This is the eighth in a series of posts about how I ended up where I am today. Next: the worldwide role, the shift from classic ML to generative AI, and watching the acceleration up close.