I learned to ask what happened before accepting what the record called it.

I joined Tervanex Components in August 1990 after completing industrial statistics at the fictional Neraveth Institute of Industrial Statistics. My first appointment was within a twelve-person analysis team. I contributed analysis within that team; colleagues and expenditure remained under my manager’s authority. The company was much smaller than it is in this invented present-day record, and a question about a production figure could often be taken directly to someone who knew how the record had been made. That access taught me to look beyond a column heading.

Factory records were practical documents before they became material for enterprise analysis. A code might be perfectly clear to the people using it on one site, while carrying a different meaning elsewhere. I began by helping colleagues examine those distinctions. My early work did not amount to authority over production, quality or engineering. It gave me experience of asking how an apparently straightforward total connected with the physical work behind it, and of hearing explanations that were absent from the data itself.

In April 1997 I became master-data supervisor. By 2003 the remit covered twenty-one colleagues through three direct reports, with S$950,000 in annual function expense. The work required more than imposing a consistent naming convention. A current part definition, an alias and a superseded definition had different uses, and removing one without understanding its relationships could make earlier records harder to interpret. I wanted a common account that remained intelligible to the people who had to use it, rather than a short catalogue whose apparent tidiness concealed lost context.

I moved into plant analytics in June 2004. In 2008 I endorsed a combined lot report that treated eighty-six codes as eighty-six physical production lots. Mei Arvellan challenged the result using plant records. Twenty-six codes had been used in two plants and sixty in one, so the population contained 112 lots. I had let the identifier stand in for the physical entity. We withdrew the claim of complete traceability and incurred S$640,000 in review expense. Quality colleagues determined what operational examination was required; I could not correct my analytical error by assuming their authority.

The vice-president appointment in August 2012 expanded my responsibility across industrial data. The harder questions concerned what a combined dataset allowed us to claim. A 2017 maintenance-model evaluation drew records from forty-two assets into random row-level splits. That arrangement did not establish performance on unfamiliar assets merely because the evaluation rows had been held apart. I asked for the claim to be withheld and for the intended use to determine a more appropriate evaluation. S$2.3 million in further analytical work remained a cost, without a later success result added to make the account more comfortable.

I became CDO in March 2019 and now report to the CEO. My 158-person function works through seven direct reports; the company's 6700 employees and S$2.4 billion revenue describe the institution around us. My S$21.4 million operating-expense delegation funds the data function. It does not give me control of the factories' capital budgets, machine settings or product decisions. Those distinctions become especially important when a data programme crosses several sites and its findings affect people outside the team that prepared them.

Recent work has examined clocks, part histories and supplier submissions. Thirty-seven of two hundred machines in a 2021 review showed observed clock offsets, making the apparent sequence of events open to question. The 2023 parts programme separated current definitions, aliases and superseded definitions instead of treating every additional record as waste. In 2025 supplier footprint submissions required measured, modelled and missing information to remain distinct. These were different problems, linked by the risk of a convenient representation becoming stronger evidence than the underlying record supported.

I have presented to the board as an executive since March 2019 and have not held a statutory seat. Preparing for a first appointment would mean learning the institution's responsibilities and considering where my experience complements its existing expertise. I would bring questions about the basis of a claim, together with the memory of having made one that was too strong. Tervanex, its people and these results are fictional. The purpose of this profile is to show how an industrial data leader can describe judgement, scale and mistakes without allowing a polished account to obscure what the records actually establish.