Full-stackRamco SystemsApr 2026 to Aug 2026
Automated configuration audits for global payroll
Payroll consultants checked each customer's configuration by hand, one SQL query and one spreadsheet at a time. I built a platform that turns their checks into reusable rules and runs them on a schedule across every customer and country.
My roleSole developer, with input from the payroll experts
Large audits about 8x faster, and every finding traceable to the rule, the data and the reviewer.
Built at Ramco Systems. Internal names and customer data stay private.
The problem
The payroll centre of excellence checks that every customer’s payroll configuration is correct: statutory setup, pay elements, organisation structure. That meant hand-written SQL and spreadsheets, with no record of which checks ran against which data.
What I built
A platform where a check is written once (or drafted by AI from plain English) and then runs automatically against every customer, country and operating unit through a read-only API. Failures go to a review queue with the failing rows and a full audit trail.
Hard parts
- Complete data, fast. Full-data pulls fetch pages and checklists in parallel under one global cap on concurrent requests, with retries. Large pulls went from about 15 to 18 minutes to about 2, with identical findings.
- Country-correct results. A rule for one country only runs against the operating units registered to it, read from each customer’s own configuration.
- AI that can’t invent columns. The model sees the real column list and must answer in a strict schema; invalid output gets up to three self-corrections, and the result is always a draft a person approves.
- Python
- FastAPI
- React
- TypeScript
- SAML SSO