I build software around the way operations actually work.
I work with the people running a process, find where records and decisions break down, and build the right control layer: a spreadsheet system, an integration, or internal software.
The archived system flags the original transaction as VOID and creates an offsetting row. A later review found that a second void could target a reversal row. The hardened reconstruction excludes reversal rows and checks the resulting balance.
Inventory Control fictional data
ItemMovementBalanceStatus
Safety Gloves−48LOW
Router−13VOID
Router+14ACTIVE
ITEM_LISTLEDGERDASHBOARD
1 / 3
The work hiding between tools
Small gaps make everyday operations hard to trust.
01
Records without history
A balance exists, but nobody can quickly explain what changed.
02
Work between systems
Reports arrive in Slack and email, while follow-up stays manual.
03
Exceptions in memory
Low stock or an invalid submission is visible only to the person checking.
How I work
Understand the operation. Then choose the technical layer.
MapPeople, inputs, handoffs, decisions, and exceptions.
ProveThe risky logic: balances, reconciliation, validation, or control.
FitThe smallest system the team can use and maintain.
Excel / Sheets→Apps Script + integrations→Internal application→Full product system
Choosing the delivery layer
A spreadsheet can be the interface. The workflow is the real product.
Excel or Google Sheets can be right for structured entry, reconciliation, summaries, and visible exceptions. Apps Script adds validation, alerts, and integrations.
When the process needs stronger permissions, concurrency, or scale, I carry the same workflow understanding into React, Next.js, Node.js, APIs, PostgreSQL, real-time services, and payments.