Product thinking,
operational depth.

The problem

When I took ownership, data accuracy sat at ~50%. Multiple teams entering records with no validation rules, no standardised formats, no accountability checkpoints. Errors contaminated weeks of reporting before surfacing.

Leadership was making decisions on data they couldn't trust, with no visibility into where errors originated.

My approach

1. Mapped the full data flow from entry to reporting, identifying seven handoff points where validation should occur
2. Designed an AI-assisted audit workflow using OCR and response ID cross-verification to catch discrepancies at the source
3. Built automated flagging rules that routed errors to responsible teams before downstream reporting
4. Created a structured resubmission process so corrections were traceable and didn't reintroduce new errors
5. Established a real-time compliance dashboard to give leadership ongoing visibility into data health

The result

Data accuracy lifted from ~50% to 97%. Manual correction hours largely eliminated. Executive reporting now runs on clean, trusted data: 5,000+ records per month with minimal overhead.

The problem

Operational findings (non-compliance, safety concerns) were collected through Google Forms — external, passive, no notifications. Submissions were reviewed only weekly or biweekly.

A serious finding logged Monday might not be escalated until the following week. No notification, no submission visibility, no accountability checkpoint.

My approach

1. Identified the gap: passive external form with no escalation trigger and no admin notification
2. Designed an in-app findings submission flow integrated into the session end workflow; users log findings before leaving the session screen, not in a separate tool
3. Wrote functional specifications and logic flows for the submission form, notification trigger, and the admin review panel within the in-house app's admin view
4. Coordinated with the outsourced development team across multiple briefing and feedback rounds to build the feature
5. Ran UAT across submission scenarios, notification triggers, and the admin escalation interface
6. Released with user communications, in-app guidance, and post-launch adoption monitoring

The result

Admins receive a real-time notification the moment a finding is submitted. Issues escalated within 48 hours, down from a passive weekly cycle. The findings workflow now lives entirely inside the operational app.

The problem

Monthly Airwallex disbursements for 5,000+ coaches, facilitators, and auditors required 240+ min of manual work per cycle: cross-referencing invoices, applying deductions, reconciling discrepancies, and building the file from scratch.

No standardised format, error-prone, entirely dependent on manual checking before disbursement.

My approach

1. Audited the full payment workflow: invoice verification, deduction application, reconciliation, and file generation
2. Built an advanced Excel payment automation workbook with structured data pipelines, formula-driven deduction logic, and automatic disbursement file generation
3. Designed the workbook to handle KPI incentive calculations, standard payments, and transportation reimbursements across different user groups
4. Built in validation rules that flag discrepancies before the file is finalised, eliminating the post-hoc error discovery cycle
5. Documented the process as an SOP so the workflow could run consistently regardless of who operates it

The result

Disbursement file prep dropped from 240+ min to under 5 min. Accuracy and standardisation improved. The Excel automation framework was later adapted for the HWE payment system covering KPI incentives and transportation reimbursements.