mysu.
Index // Selected work

Proved
outcomes.

Selected work  //  05 projects
01JSI Ledger02themysu.com03Harbr04ResGo05Capilingua
01 / Engineering

JSI Ledger

0
weeks from brief to live

The problem

Everything was on paper, and none of it was defensible. No document numbers, so nothing could be traced or proven complete. Receipts drifted away from the payments they belonged to. Month-end meant a day of reconstructing the story for the accountant out of whatever had survived the drawer, with no way to show an auditor that what survived was all of it.

What we did

We built the controls in, rather than bolting process on top. Voucher numbers are issued by the database under a row lock, so the sequence has no gaps and no duplicates even when several people record at once. Every voucher is issued in both directors’ names and carries both signatures. The receipt photo is bound into the same PDF as the payment it evidences. Once a voucher has been sent, nothing can delete it: a mistake is voided with a reason and stays in the record, and an append-only log keeps the history. Then we automated the month: one tap collects the period, generates the Excel ledger and every voucher PDF, flags any payment still missing its receipt, and emails the complete pack to the accountant. Installs on any phone, in English and Sinhala.

The result

A day of month-end assembly became one tap. The company now has something it did not have before: a complete, numbered, evidenced record it can put in front of an accountant or an auditor and stand behind.

02 / DTC & Commerce

themysu.com

0
ways to pay, one checkout

The problem

Selling internationally from Sri Lanka means local payment methods that no off-the-shelf checkout supports together, and a content model the team can actually edit.

What we did

A storefront on a stack we own end to end, with four payment methods behind a single checkout and content the team edits themselves, with no developer in the loop to change a product page.

The result

Our own label, our own revenue. The architecture is tested on us before it is ever recommended to a client.

03 / Data & Audit

Harbr

0
months to launch

The problem

Reporting was assembled by hand from several sources. Two people could answer the same question differently, and nobody could say which was right.

What we did

A full-stack build, interface and data layer both, with tracking wired at the source so the reported figure and the real one are the same number.

The result

Two months from brief to launch, with one set of figures everyone works from.

04 / Engineering

ResGo

0%
reduction in no-shows

The problem

For most restaurants, reservations are still fragmented: a phone ringing through service, a paper diary at the host stand, and marketplace platforms that charge per cover and keep the guest relationship for themselves.

What we did

We operated as ResGo's product and engineering team. We shaped the proposition with the founders, designed the guest and operator experiences, built the platform, and carried it through to launch. One accountable team across the entire arc.

The result

Live in production and serving restaurants, on an architecture designed to grow from a single venue to many without rework. More than 7,000 guests seated through it, including at award-winning venues.

05 / Engineering

Capilingua

0
extra admin staff, even as intake grew

The problem

As the school grew, friction compounded. Course administration lived across spreadsheets, enrolment was manual, and every new intake added overhead that scaled with headcount rather than with software. Off-the-shelf platforms forced them into a generic mould.

What we did

We worked directly with leadership and teaching staff, translating pedagogy into product workflows: course and cohort management aligned to level progression, lesson materials organised the way teachers actually plan, and separate experiences for students, teachers and admins.

The result

Capilingua now runs courses, assessment and administration on one platform built for its own way of teaching. Administrative overhead no longer grows with every intake, so they can expand without adding headcount.