Settlement + Reconciliation Engine: 99.95% Accuracy at $1B+ GTV
A multi-rail settlement and reconciliation engine, canonical double-entry ledger, three-way auto-reconciliation, exception management and corridor-aware payout windows. Closed the gap between treasury, finance and product at $1B+ GTV.
What this is, in one paragraph.
Replaced spreadsheet reconciliation across five rails with a canonical ledger, three-way auto-recon and an exception taxonomy that fed product. Made unit economics observable per rail, per corridor, per merchant.
Rail events, internal ledger and bank statements are matched continuously. Exceptions are typed and routed; the defect stream feeds product backlog so the same break never recurs.
The job to be done.
Multi-rail flows across cards, wallets, IBFT, DCB and cross-border corridors created reconciliation breakage that finance and treasury were absorbing manually, slowing payouts and obscuring real margin.
What I shipped.
- Canonical transaction ledger across all rails
- Automated three-way reconciliation: gateway, bank/partner, internal ledger
- Settlement scheduler with corridor-aware payout windows
- Exception workflows with root-cause tagging fed back to product
How it's put together.
- Double-entry ledger as the source of truth; rails publish events that post entries
- Recon as a stream join across three sources with a tagged exception store
- Settlement scheduler reads available balance per merchant per currency per corridor
How it actually runs.
- Finance signs off on the ledger model and exception taxonomy
- Every exception type has a product owner and a recovery SLA
Where I sat in the work.
Product owner working alongside finance, treasury and engineering. Defined the ledger model, exception taxonomy and settlement SLAs.
What moved.
- 99.95% reconciliation accuracy at $1B+ GTV scale
- Eliminated manual journal entries for core flows
- Made unit economics observable per rail, per corridor, per merchant
What I chose against.
- Held back a faster payout window until recon confidence was demonstrable
- Built our own exception store rather than buying a recon tool, better feedback loop, more ownership
What I'd take into the next build.
- Settlement and reconciliation are not back-office problems, they decide trust with merchants and partners.
- If finance is your reconciliation system, you don't have one.
Relevance to networks, PSPs and cross-border platforms.
The transferable core: a canonical ledger, three-way reconciliation and an exception taxonomy that make settlement observable per rail and per corridor. Acceptance, scheme settlement and treasury teams are recon products at heart, inside every network, PSP and cross-border platform; only the logo changes.
Discussing payment infrastructure / product leadership roles?
Reference-available. Download the résumé or get in touch.
Adjacent builds in the same operating context.
Simpaisa Payment Infrastructure Platform
A regulated, multi-rail payments platform processing $1B+ annual GTV and 270M+ payments a year across pay-in, payout, wallets (DCB/IBFT), card...
Merchant Onboarding + KYC/KYB Automation
Automated merchant onboarding pipeline, KYC/KYB, UBO discovery, sanctions and PEP screening, risk-tiered decisioning with full audit trail....
Fraud, Risk and AML/CFT Controls: Layered Decisioning at $1B+ GTV
Layered fraud, AML/CFT and sanctions decisioning built natively into the payments stack, vendor signals, device intelligence, internal velocity...