← All work

Data quality & security · Xylem Learning

Making the numbers finance relies on actually add up

Reconciled payments across systems: traced missing payments and pushed them through, then automated the App to CRM to ERP check as a Metabase dashboard.

Role
Analyst, then owner of the data layer
When
2024 – present
Stack
SQL · Metabase · Power BI · LeadSquared CRM · Python (pandas)

150+

SQL reports and data pulls

30+

record types locked down

1,100+

bulk data imports run

The problem

Payments travel from the app to the CRM to the ERP, and some went missing on the way. Finance reports are only as good as the weakest hand-off, and student records needed tighter control over who could export them.

What I did

  • Reconciled payments across App, CRM (LeadSquared) and ERP: traced payments missing over six months and pushed them through, imported missing app purchases, flagged payments stuck at the applicant stage, then automated the check in Metabase.
  • Caught CRM-to-ERP data issues (duplicate applicants, missing lead IDs, unrecorded fee payments) and drove the record fixes.
  • Wrote root-cause analyses on CRM data lost between applicant and enrollment and on recurring data corrections, and reworked the LeadSquared lead-scoring model.
  • Designed a fee-sanity check that reconciles paid amounts against fee, transfer and refund records, with one-click correction and an audit trail.
  • Tightened access control: removed custom permissions from 30+ record types, export and print rights on sensitive records, and masked personal data server-side.
  • Mapped the ERP's personal-data flows to third-party vendors (LeadSquared, MSG91, Gallabox, Razorpay) for a product data questionnaire.
  • Built 150+ SQL reports and data pulls in the ERP, including fee approval, hostel capacity, fee balance and payment sanity reports used by finance and centre teams.
  • Built analytics dashboards with drill-down for enrollments, revenue, hostel occupancy, discounts, refunds, attendance and exam results, reconciling centre and vertical KPI definitions.

Outcome

Built a fee-sanity check with one-click correction and an audit trail, catching paid-amount mismatches before they reach finance reports.

Locked down student data: removed export and print rights on sensitive records and masked personal data in self-serve reports.