Workforce Integrity Engine

Continuous payroll-integrity intelligence with human-controlled payment intervention

Problem

Traditional payroll systems process transactions but often discover ghost workers, duplicate accounts, salary anomalies, and stale identity records only after funds have been disbursed.

Workforce Integrity Engine was designed as an intelligence layer above existing payroll infrastructure, helping institutional finance and audit teams identify suspicious behavior before a payment leaves.

Solution

The prototype correlates payroll records, attendance behavior, verification freshness, historical trust patterns, peer-group deviations, and payout anomalies.

It demonstrates an operational review workflow designed to:

Generate evolving employee trust scores
Surface explainable anomaly evidence
Identify suspicious relationships such as shared payout accounts
Route high-risk records to human investigators
Pause, approve, or release payments through Squad-backed workflows
Preserve an audit history of model signals and human decisions

My Contribution

I worked as the Frontend Engineer in a four-person team.

My responsibilities included:

Building the Next.js interface for payroll review and risk investigation
Presenting multi-signal risk scores without reducing decisions to opaque AI labels
Implementing workflows for anomaly evidence, trust monitoring, and payment intervention
Connecting frontend journeys to project APIs and Squad's test environment
Collaborating with backend and machine-learning engineers during rapid hackathon delivery

The wider team delivered the backend, intelligence models, data workflows, and supporting product work. The hackathon result belongs to the team; the responsibilities above describe my personal contribution.

Product and Engineering Approach

Explainable Risk Intelligence

Deterministic fraud rules detect conditions such as duplicate payout accounts and unusual salary changes
Statistical anomaly detection surfaces deviations from historical and peer-group behavior
Every flagged record includes evidence and reasoning for investigator review

Human-in-the-Loop Controls

The prototype does not make irreversible employment or financial decisions autonomously. Its demo flow requires investigators to review the evidence before approving, escalating, blocking, or releasing a disbursement.

Squad Integration

The prototype connects its frontend workflows to Squad APIs and a virtual-account test environment for identity checks, payout controls, and transaction-linked audit flows. In the demo, payment release follows the relevant human-review step.

Frontend Stack

Next.js and TypeScript
Tailwind CSS
TanStack Query for server-state workflows
Zustand for focused client state
Recharts for operational risk visualisation

Outcome and Recognition

The team delivered a public working prototype and reports advancing to the top-20 team stage of Squad Hackathon 3.0. Independent event coverage reported more than 1,600 undergraduate applicants; the top-20 figure describes the team's competition stage rather than an individual ranking.

Why This Project Matters

Workforce Integrity Engine demonstrates frontend engineering for a high-stakes operational system: dense data, explainable machine-learning signals, financial controls, audit requirements, and human judgment all need to work together clearly.