The explanation that should travel with the code

One of my earliest maintenance assignments concerned a program whose original developer had left. The source was available and the routine still ran, but nobody could explain why a particular boundary condition received different treatment. The easy answer was to preserve it indefinitely. The equally easy alternative was to replace it with something that looked more consistent. Neither answered the question a responsible maintainer needed to resolve: what obligation had the original behaviour been intended to fulfil? That experience made explanation part of my idea of engineering quality.

I completed a computing degree at Renswick College in May 1992 and joined Delvarion Software in June. I worked in a nine-person development team without direct reports or independent expenditure authority. We built and maintained applications for commercial customers. I learnt to reproduce a reported problem, distinguish an intended rule from an accidental implementation and leave a useful account for the next person assigned to the code. By the end of 1999, the company employed 240 people and reported USD32 million annual revenue. Those figures describe the employer, not a business I personally managed.

I entered Talverin Bank as engineering manager in January 2000. Banking made the consequences of seemingly minor software choices more tangible. A technical correction could affect a downstream application or a business interpretation even when an individual function looked better in isolation. I reported to the head of applications and led twenty-four engineers through three reports by the final year of that appointment. My 2007 authority was USD4.6 million. The bank then had twenty-two branches, 900 employees, USD8.2 billion assets and USD310 million annual operating income.

The head of platform engineering appointment began in January 2008. I became responsible for common technical services and the engineering practices that allowed application teams to use them. By 2014 I led sixty-one employees through four reports, with USD17 million annual authority. The bank had grown to thirty-four branches and 1,400 employees, with USD15.6 billion assets and USD590 million operating income. I learnt that a shared component transfers responsibility as well as capability. If its behaviour is poorly described, every receiving team pays again to discover its limits.

I became engineering director in January 2015, reporting to the CTO. My team reached 112 employees through five direct reports, with USD39 million authority in 2019. The bank's forty-four branches and 1,900 employees then operated within a USD24.2 billion asset base and USD890 million annual operating income. I worked on the relationship between software delivery, maintenance and architecture. The useful question was often not whether a team could build a new service, but whether its successors could safely understand, change and retire it.

My CTO appointment followed in January 2020. I report to Leona Wexmere, the CIO, and lead 162 employees through five direct reports, with forty-two contractors counted separately. The 2026 USD62 million authority covers USD38 million platform engineering and USD24 million maintenance. Talverin remains an unlisted regional bank owned by institutional investors and founding families. Its December 2025 USD31.8 billion assets, USD20.4 billion loans and USD25.6 billion deposits describe the enterprise context. I do not claim credit or balance-sheet authority from my proximity to the systems supporting that business.

My recent decisions have made incomplete transitions explicit. Two older consumers retained their existing contract after a compatibility programme. Two release artefacts remained held because their build histories could not yet be supported. Twelve synthetic rounding examples stayed unresolved after a calculation rehearsal. In each instance I preferred a bounded technical account to a claim of completion that would require a receiving team to rediscover the unanswered question. The remaining work had a cost and an owner; it did not disappear because most of the examined population had passed.

I also challenge how we describe a platform's value. Nine early adopter teams saw the median interval from approved request to test environment fall from 4.2 to 2.1 days. That was an observation about the same cohort and a particular stage of work, not evidence that production releases doubled or that engineers could be removed. Two further teams adopted after support work, while one retained a justified specialist route. I want the common platform to earn use through a suitable service, with its exceptions governed openly.

Since July 2023 I have separately served as a non-executive director of Averlisle Learning, a private adult-training enterprise. That concurrent role has twelve annual days allocated and does not make me a Talverin director. It has taught me to contribute through collective oversight rather than carry my executive operating habits into another management team. Any further appointment would need employer agreement, relationship review and capacity for eighteen additional days within a thirty-day external envelope. I would bring experience in the durability of software decisions, and a willingness to learn the commercial questions those decisions must serve.