Why financial AI built by someone who lived it works differently.

In 2017, I joined State Street Global Advisory as a Reconciliation Analyst. Every day: Bloomberg Terminal, Refinitiv Eikon, SmartStream TLM. Reconciling 100+ managed pension fund portfolios — equities, derivatives, FX, post-trade settlement, corporate actions. The grinding, precise, high-stakes work of institutional asset management.

Most people would see that as a detour before a "real" tech career. I see it as the most important thing I ever did. Because when I crossed over to data science at LSEG in 2019, I didn't just bring Python and statistics — I brought an intimate understanding of the data I was now being asked to automate.

I knew why Korean regulatory filings were structured the way they were. I knew what made a broker report ambiguous. I knew which edge cases would break a production system in the third month, not the third day. That knowledge is not in any dataset. You get it by living inside the domain.

Nine years and £600K+ in delivered value later, the same pattern keeps repeating: the best financial AI comes from teams that combine deep market expertise with serious engineering. I've done both sides. Now I'm looking for the next stage in teams building production AI for financial markets — whether that's fintechs, exchanges, asset managers, or financial-services arms of global technology companies.