Debt Collection Management Platform

The platform is built as three coordinated applications. The client app is where creditor businesses (and their invited staff, across multiple divisions/business units) register, submit new debt cases individually or in bulk via XLSX upload, track each case's status and financials, exchange messages with the agency through a task system, and pull reports on their portfolio.

The admin app is the agency's operating console: staff review incoming cases and client rosters, move cases through a configurable status/sub-status workflow, import bulk case files, manage collectors and internal users, inspect audit logs, recover soft-deleted records, and run cross-client analytics and reports.

Both apps talk to a single NestJS backend that owns the domain model — cases, clients, divisions, tasks, statuses, users — enforces role-based access so a client only ever sees its own data and handles the heavier lifting: parsing bilingual (Dutch/English) XLSX imports, generating and emailing XLSX exports, storing attachments in S3, and keeping an audit trail of every change.

A case itself carries the full lifecycle of a debt: debtor identity and address, the collector assigned, principal amount, interest, collection fees and total owed, amount recovered so far, currency, age in months, and a status/sub-status pair — with client comments kept separate from agency-only fields so updates never overwrite each other.

The client is a Europe-based B2B financial services provider specializing in debt collection and Accounts Receivable management. They help businesses recover outstanding and overdue payments while focusing on maintaining strong customer relationships. Their services include Debt Collection, A/R Management, Training, Consultancy, and Secondment, with support for businesses operating across international markets.

• Dual-Portal Data Isolation: Designing a single backend API that serves two distinct frontends while ensuring strict multi-tenant data isolation. The system needed robust Role-Based Access Control (RBAC) so clients only see their own portfolio, while admins have a global view.
• Complex Data Ingestion: Building a reliable, bilingual (Dutch/English) XLSX import engine. Debt cases come in varying formats, requiring dynamic header mapping, row-by-row validation, and intelligent create-or-update logic based on client and reference numbers.
• Configurable Workflows: Instead of hardcoding case states, the system required a flexible status and sub-status workflow that administrators could configure and manage as master data, accommodating different collection strategies.
• Data Integrity and Auditability: In the financial sector, tracking changes is critical. Implementing a comprehensive audit trail, case-history tracking, and recoverable soft-deletes required careful database design to avoid performance bottlenecks while maintaining a complete system of record.

A three-tier architecture: Two independents React single-page applications (client portal, admin portal) consuming one shared REST API, with PostgreSQL as the system of record.

The API is built on NestJS with a modular structure — separate modules for auth, cases, clients, tasks, imports, reports, audit, currency, master data, settings, storage, and scheduled cleanup — each with its own service, controller, and DTOs, documented automatically via Swagger.

Authentication is JWT-based with role guards enforcing client-vs-admin and per-client data isolation at the query level. Case data enters the system either through a bilingual XLSX import pipeline (matching rows to existing cases by client + internal reference number, updating or creating as needed) or through the client's own case-submission form, which is converted back into an XLSX and emailed to the agency inbox via SMTP/AWS SES. Uploaded files are stored in AWS S3. Every case change is written to an audit log and a case-history record, soft deletes are recoverable from a dedicated admin screen, and a scheduled job handles routine data cleanup.

The whole backend ships as a Docker image, deployed automatically to AWS ECS on every push to main via a GitHub Actions pipeline (build → push to ECR → render task definition → roll out to the ECS service).

Backend: NestJS, TypeORM, PostgreSQL, JWT / Passport, AWS SDK, Nodemailer, Multer, SheetJS (xlsx), Swagger, class-validator

Frontend (Client + Admin apps): React 18, TypeScript, Vite, Redux Toolkit, React Router, TailwindCSS, shadcn/ui (Radix UI), Recharts, ExcelJS, Zod

• Single System of Record: Eliminated the reliance on fragmented spreadsheets and email threads, centralizing all debt case data, financials, and communications in one secure platform.
• Operational Efficiency: The admin portal streamlined the agency's workflow, making bulk imports, case tracking, and reporting significantly faster and less prone to manual errors.
• Enhanced Client Transparency: Creditors now have real-time visibility into their cases, recovery progress, and financials through the self-serve client portal, reducing the need for manual status updates.
• Improved Communication: The built-in per-case task thread system directly connected the agency's staff with client representatives, speeding up query resolution and document exchange.

The Multi-Tenant Debt Collection & Recovery Management Platform successfully transformed a manual, disconnected process into a streamlined, cloud-native workflow. By providing tailored, dedicated portals for both the collection agency and their clients—backed by a robust, secure, and unified backend—the project not only improved operational efficiency and data accuracy but also significantly elevated the client experience.