BGV Employee Background Verification Platform

Overview

I am the only engineer on a multi-tenant background-verification (BGV) SaaS platform used by an India-based verification-services company. It manages the full lifecycle of employee and business background checks — corporate clients onboard people to be vetted, 14 verification types run per subject, field work is farmed out to external vendors and law firms, and results roll up into a colour-coded case outcome with downloadable reports.

I designed the architecture, wrote the NestJS API and the Angular SPA, own the database and the deployment, and handle every ticket end to end. The codebase is roughly 105,000 hand-written lines across the two repositories and has been in continuous production use for several years.

Executive Summary

The platform is a mature, multi-tenant B2B application that manages the full lifecycle of employee and business background checks: corporate clients onboard people (or companies) to be vetted, 14 different verification types are run per subject, field work is farmed out to external vendors and law firms, and results roll up into a colour-coded case outcome with downloadable reports.

It has been in continuous production use for several years and, from database snapshots, currently serves roughly 40 corporate clients, ~3,600 user accounts, and ~3,600 candidate cases, backed by 12,000+ uploaded verification documents.

I am its only engineer. I designed the architecture, wrote the backend and the frontend, own the database and the deployment, and handle every ticket end to end — from reproducing a client's prose bug report, through the fix, to sign-off on scope.

The Actors

Lifecycle of 14 verification types

Each check is its own self-contained module — entity, controller, service, export service, repository, mappers, DTOs — generated from one blueprint:

Reporting

• MIS spreadsheets — Excel workbooks (one per verification type, plus a candidate-level roll-up) clients use to track a batch of cases and their statuses. Built with a spreadsheet library, streamed as downloads.
• Per-candidate PDF report — the formal deliverable, compiled from templates and rendered to PDF with headless Chrome, with per-section partials for each verification type.

System architecture

Backend — a disciplined modular monolith

Every feature is a NestJS module with the same internal layering: Controller → Service → Repository (a custom class per aggregate) → MySQL, with dedicated Mapper classes translating between DTOs and entities. Side effects — email, notifications, heavy row processing — leave the request path through an event bus and six Redis-backed Bull queues, each with a processor that logs queue lifecycle events and marks failures.

Frontend — Angular + NgRx at scale

Technology stack

Backend

Frontend

What this demonstrates

• Backend engineering — an 18-module NestJS monolith with consistent layering, an event-driven side-effect architecture, six queue processors, and a deliberately-tuned posture for heavy synchronous work.
• Frontend engineering — a ~130-screen Angular SPA with ~32 NgRx feature stores, disciplined effect patterns, three-layer RBAC, and genuinely complex nested dynamic forms.
• Database — owning a ~68-entity MySQL schema; writing migrations with backfill and rollback; recognising and reporting schema drift.
• Security — finding and closing a cross-tenant data-exposure gap; finding unauthenticated admin APIs; reporting committed-secret exposure with a remediation plan.
• Debugging — root-causing a 100%-failure bulk pipeline through live reproduction and SQL inspection; a cross-domain date/timezone bug; a latent async race; a latent crash.
• DevOps — Docker and serverless deployment paths, S3 static hosting, environment management.
• Judgement — separating the ticket from the system state, fixing at the right layer, agreeing scope in writing, and reporting honest negative findings.