2016
Just before my 16th birthday, my brother asked me a question:
Do you think you can build a bus booking website?
He'd just started a company with his friend to provide consultancy for school bus services, assisting with network routing and scheduling. Their first client was the school I was still attending (he had just graduated), and they wanted to digitise their paper booking form. In our brave new world of LLMs this type of thing would be trivial for anyone to produce in an afternoon, but a decade ago they had to seek out the next best thing: a teenager with a passion for coding.
I read everything I could find online, settled on a LAMP stack (Linux + Apache + MySQL + PHP), and put together the first version in a few months:
We then began a working arrangement which would continue indefinitely, where my 2 bosses would promise functionality to people and I would figure out how to deliver it. At first I found this deeply frustrating, as I felt an enormous pressure to create the impossible, but over time it developed in me a certainty that I can figure out basically anything.
2017
We started gaining clients, and soon I was maintaining 5 copies of roughly the same codebase, with minor tweaks for each school. I realised this was untenable, and decided to create a single system which could manage as many websites as we wanted, with different features available with settings.
Not a groundbreaking concept, now that every man and his dog has built a SaaS, but this was almost 10 years ago, and I was 16, so I was quite pleased with the idea. Pre-LLM, finding the right technical word to search for was sometimes the key to a good solution, and here the word was 'multitenancy'.
The result was that the marginal developer time cost of doubling the number of sites over the next year was basically zero!
This was the beginning of another learning: you Are gonna need it (yAgni); a direct contradiction of the common coding wisdom that you aren't gonna need it. I completely agree that you shouldn't create features that you don't know your users need. But foreseeing a need, particularly needs relating to scaling, and particularly when you have full oversight and control of a project, can save a lot of time and effort! It's a subtle art.
Of course this was a double-edged sword for me. I had reduced the work required for me to keep these sites running, which meant I was being paid for fewer hours. To make my goals align with the company's I realised that I should get a slice of the pie, and so I entered into my first serious business negotiation.
It was tough, but I had my grandfather as legal counsel, and we came to a revenue share arrangement that meant that I was invested in making a high quality sustainable product.
2018
A year or two in, we started doing some consultancy for some real bus companies. I was tasked with pulling as much data as possible out of their ticketing system (I was contacted by their IT team and told off for overloading them), and answering various different questions using the data. In the 3 days leading up to the biggest deadline, I put my bed next to my desk, and would start analysis running, nap until it was done, and then start the next stage. I don't think I slept more than 30 minutes continuously for the whole 72 hours!
I'm pretty sure I didn't mind; it was interesting work, it made me feel very important, and I didn't really need much sleep. The joys of being young.
I also remember sitting in my parents' kitchen stuffing about 500 envelopes twice (we messed up the first time), to send to the pupils of our biggest client at the time, and it really felt like we were having the startup experience™️.
2019
As the system grew, and I approached my A levels, it became clear that we were going to need more dev resource than just me. So in mid-2019 we hired a placement student for a year. This turned out to be an excellent decision, as he came back full-time after finishing university, and grew into a fantastic developer, providing invaluable leadership and mentorship to the team.
2020
My main design skill lies in user experience, not visual design, but I've definitely developed a strong sense of what looks good to me[1], and I started to feel that our logo wasn't up to scratch for the company that I saw us becoming. So I managed to convince the master of coin that we should spend (what seemed at the time) an enormous sum of money on having one professionally produced.
We hired a real design firm, and created brand guidelines which were rolled out across all of our marketing, and are still in use today!
It's impossible to tell whether anyone was ever put off by our old logo, or convinced by our new one, but in my eyes it was a big step in the professionalisation of our growing company. And I think that one of the most important skills that I've plied over the years is to fight for what I believe in; pushing for ideas that I felt were valuable, and not being steamrolled into accepting ones which I felt were misguided, even by people more senior than me.
2021
Alongside all this, the company had branched out into running public service vehicles on the roads. Coach hire, bus services, rail replacement, you name it. As this fleet grew, its management became more complex, and so we started creating software to assist. This was a major step-change from the relative simplicity of 5-10 routes, all centring on a school, running just twice a day, and over time the system was involved in everything from payroll to vehicle maintenance.
2022-2024
My team of developers grew to 7, the customer service team grew to more than 10, and our client list grew to 150. We built functionality to help schools with managing sports trips, allowed parents to track the location of the bus through an app, and created deep links with other school management software.
I stepped into my absolute favourite role that I've had in my career so far: greasing the wheels. My team worked with schools to fulfil their needs, and I worked to fulfil my team's, by improving processes and introducing automation. We massively increased test coverage, and started type-checking, linting, and auto-formatting. We added a CDN, caches, auto-scaling, infrastructure as code, and instrumentation to provide insights into our running systems.
All of these things are completely standard in industry, but this codebase was essentially my learning environment, and so every 'best practice' I rolled out felt like another step on my journey to becoming a professional developer.
2025
Last summer, it became clear that our two diverging arms (school transport software, and our internal fleet) were splitting people's attention, and so the schools business was sold to an EdTech company. It was sad to say goodbye to my team, but it was great to know that this thing that I'd been building for so long was going to be well looked after.
2026
I kept working on the software for the bus company, and got to work with a fantastic developer, who brought in a radical new suite of functionality, to support the needs of our growing fleet.
I also started a two year Master's in Quantum Computer Science in Amsterdam, so again my attention was split.
What now?
At the half-way point of my degree, with my thesis project looming, I realised that I couldn't keep all the plates spinning. I've been managing the development of software in various capacities alongside my GCSEs, A levels, and Bachelor's degree, but it seems that a Master's is just too much. As such, after over 10 years, it is time to say goodbye to the company, and all the wonderful people that I've worked with along the way.
I'm very excited to now apply all these years of experience to my next projects, starting with an internship at a quantum simulation company, followed by my Master's thesis working with formal proof assistants.
My life would be vastly different without this opportunity, and I cannot imagine the trust it took to offer it to my 15 year old self. I cannot thank Peter and Dominic enough.