Tiger Analytics, a global leader in data science, partnered with 18North to deliver a digital web platform that wraps around a machine learning–based optimizer model used for planning retail field force deployments in the Australian market.

While Tiger Analytics developed the ML-based Optimizer, which mimicked and improved upon the legacy Excel planning model used by the end customer, 18North was responsible for building the frontend and backend application layer. This wrapper allowed structured ingestion of planning inputs, real-time validations, orchestration of the Optimizer engine, and intuitive rendering of its output to end users.

Deployed as a responsive, full-stack web application on Azure, this platform enabled field sales planners to simulate manpower plans across regions, activities, and banners with significant efficiency gains.

The end customer — a leading retail organization in Australia — previously used an Excel-based manpower planning tool that helped allocate sales staff to store activities based on multiple variables:

• Market calendars

• Core vs. promotional activities

• Manpower availability

• Budget constraints

• Call plans

To scale and optimize this manual approach, Tiger Analytics was engaged to build a machine learning–driven Optimizer, while 18North was brought in to deliver a robust, business-friendly digital interface — one that would manage the inputs, trigger the Optimizer model, and surface meaningful outputs for planners.

• Co-Execution Across Teams: Needed seamless coordination between Tiger’s ML development and 18North’s application architecture.

• Multiple Interdependent Inputs: Five separate structured files (activity grid, manpower supply, call plan, calendar, sales budget) required validation and harmonization.

• Reusable Output Rendering: Visit plans generated by the Optimizer needed to be dynamically visualized by banner, region, and week.

• Zero Model Rewrites: The ML model was to be treated as a black-box — no internal changes allowed; all orchestration and API wrapping had to respect this constraint.

• Scalable UI/UX: Replace spreadsheet workflows with a responsive, collaborative, browser-native experience.
• Tight Build Timeline: 14 weeks from design to deployment.

18North developed a full-stack web platform that acted as a wrapper and orchestrator for Tiger Analytics’ Optimizer model, delivering seamless planning capability to end users.

A. Discovery & UX Co-Design

• Collaborated with Tiger and customer stakeholders to define:
  • User roles and journeys

  • Input/output expectations

  • Optimizer integration checkpoints

• Delivered:
  • Interactive wireframes and Adobe XD mockups

  • UI component library for forms, uploads, reports, dashboards

• Prototypes validated with 5 end users and revised iteratively

B. Frontend & Structured Input Handling

• Technology Stack: Angular 12, Bootstrap, HTML5, responsive design

• Built interfaces to:
  • Upload five critical inputs

  • Auto-validate data consistency and formatting

  • Modify planner-level values (e.g., target utilization %)

• Optimizer trigger button allowed users to run the engine once inputs passed validation

• Post-run, the visit plan was visualized through:
  • Filters for time period, region, banner, store

  • Summary dashboard of over/under allocation

C. Backend API Wrapping & Optimizer Orchestration

• Technology Stack: Python 3.x, Django REST Framework, MySQL

• Developed middleware to:
  • Receive and validate input files

  • Call Tiger’s Optimizer script as a batch process

  • Parse model output into structured JSON

  • Feed data to frontend for rendering

• Used Swagger documentation for API contracts

• Embedded model status tracking (e.g., “Processing”, “Completed”, “Error”)

D. Testing, Deployment & Support

• QA across devices and browsers

• Deployed on MARS Azure App Service with CI/CD pipelines

• Supported UAT in Azure QA environment

Provided triage and performance tuning post-launch

• ML Optimization System Operationalized: Tiger’s Optimizer transitioned from dev tool to production-ready solution with full business usability

• Simulation and Planning Time Cut by 40%: Planners could now test multiple resource plans with input variations and quick re-runs

• Black-Box Model, Cleanly Orchestrated: 18North’s wrapper respected ML boundaries, requiring no refactoring of Tiger’s core model

• Cross-Functional Usability: Sales, HR, and Finance all accessed role-specific dashboards
• Sustained Reuse: The wrapper’s modular design allows it to be reused in other manpower planning and capacity balancing domains

This co-created solution between Tiger Analytics and 18North demonstrates the power of combining advanced data science models with robust application engineering. By delivering a wrapper that structured inputs, managed validation, and rendered outputs at scale, 18North helped unlock the Optimizer’s full business value — enabling faster, more accurate field planning at enterprise scale.