Skip to content
Jakarta, Indonesia, UTC+7 Currently CTO at Sxored
EN ID
ALM ALM Abdul Latif Munjiat
ALM ALM
EN ID
01 / Sxored

Document intelligence for credit decisions

A hybrid pipeline of deterministic parsing and AI that turns financial documents into decisions credit analysts can act on.

Case studyEN2026-07-25
RoleChief Technology Officer
PeriodMar 2024 to Present
CustomersBanks, lenders, corporates
CertificationISO 27001
01
Problem

Credit analysts were reading PDFs by hand

Every loan file arrived as a bundle of bank statements, tax returns and financial reports in whatever format the borrower happened to have. An analyst opened each one, typed the numbers into a spreadsheet, and made a call.

That work does not scale, and it is not auditable. Two analysts reading the same statement produced two different spreads.

02
What I owned

Architecture, team, cost and trust

ArchitectureGo microservices behind Temporal workflows, with circuit breakers and backpressure designed in.
TeamHired and led eleven engineers across platform, extraction and frontend.
CostCut inference spend by routing cheap passes first and reserving the large model for exceptions.
TrustISO 27001 certification, full audit trail on every extracted field.
On numbersQuantified outcomes are pending customer approval and will be added once cleared.
Resilient systems over heroics: circuit breakers, backpressure, graceful degradation designed in rather than patched in after incidents.
04
What is next

Appraisal, then the rest of the file

Quick appraisal for property valuation shipped this quarter. The next step is closing the loop from extracted field to credit decision without a spreadsheet in the middle.

Want to know how it was built?

Talk about this work All case studies