I have been an accountant since 2016. Bank and credit operations first,
then seven years of family-office work at Direct Information FZC in RAK,
UAE. There I carried a $1B+ single-UBO portfolio, kept the books for
20 companies, and ran the intercompany reconciliation across a
40-entity group. Weekly and monthly valuations, UAE VAT, treasury, audit
coordination, and supporting the US returns attached to the group's American holdings.
In March 2026 I moved back to Kerala. My background is in accounting, and I use
technology to solve practical financial problems. As an accountant, I use it to make the
work itself faster and easier, not just tidier to report on. So I use AI to build small
private tools for the parts of the close that badly need them. Nine of them now, and more
in progress. I also read and write about where AI genuinely helps in a finance function,
and where it quietly does not.
How I work
The rigour was never the bottleneck. The tooling was.
Cut-off, accruals, intercompany, sign-off: that judgement stays with me, and I would not
hand it to a model. What I remove is the friction around it. That is also why every tool
runs entirely on your own machine: a client ledger should never have to travel to someone
else's server just to be processed. Local-only is a data-governance decision, not a
feature.