Skip to content
IC Israel Chidera
All projects
Deloitte & Touche Internal · in production

Neptune Payroll Platform

An internal payroll platform for the Tax & GES department. I built AI document extraction with Azure Document Intelligence that reads payroll documents and cut manual data entry by 60%. I also lead development of the web application and maintain it in production; the platform reduced payroll processing time by 35%.

Internal tool: no public link or screenshots. This case study stays high level, with no internal architecture or client data.

  1. Payroll documents

  2. AI document extraction

    Azure Document Intelligence

  3. Extracted data in the app

    replaces manual entry

  4. Payroll processing

Simplified illustration of the workflow. Internal system: no screenshots, data or architecture details are shown.

The problem

Payroll processing involved manual steps, including re-keying data from payroll documents by hand. That takes time and leaves room for error in work where accuracy matters.

Outcomes

less manual data entry with the AI document extraction I built
60%
reduction in payroll processing time (platform outcome)
35%
faster Largest Contentful Paint
25%

What I built

AI document extraction
Payroll documents are read by Azure Document Intelligence, and the extracted data appears in the application, replacing manual data entry.
Backend integration
The Next.js application integrates with Node.js/Express and ASP.NET APIs. I work with backend engineers on API contracts, data flow and error handling.
Performance
Improved load performance by 25% (Largest Contentful Paint) through server-side rendering, caching, code splitting and lazy loading.
Accessibility
Led accessibility improvements to bring key interfaces in line with WCAG 2.1 guidelines.

Technical decisions

  1. 01

    Automate the slowest manual step

    Extracting data from documents automatically, instead of re-keying it, removed most of the manual entry from the workflow.