Skip to main content

Practical exercise

Instruction and application
In Progress

The efficiency challenge​

In this activity, you’ll step into the role of RetailMax’s ML project team, tasked with optimising a demand forecasting model for inventory planning. The model performs well, but retraining it daily has driven up cloud costs and energy use.

You’ll redesign the training pipeline to balance accuracy, efficiency, and sustainability, choosing strategies and metrics that align with business goals.

These skills are essential for building scalable, responsible ML systems that deliver high performance while minimising waste and maximizing business impact.

Practical exercise context​

RetailMax is a global retail company using AI to optimize inventory and reduce waste across regional warehouses. The data science team has built a demand forecasting model that predicts product demand using sales data, seasonality, promotions, and regional trends. The model ensures products are stocked efficiently, reducing costs and improving availability.

Recently, retraining the model daily has driven up cloud costs and energy use. Leadership wants to maintain accuracy while cutting environmental and financial impact, making efficiency and sustainability key priorities.

Current situation​

The model meets accuracy goals, but training time and costs are rising.

  • The The operations teamwants faster retraining for market responsiveness.
  • The data science team is exploring lighter architectures and tuning methods to lower compute use.
  • The finance team aims to reduce cloud spend and improve ROI.
  • The sustainability office pushes for greener, lower-emission AI practices aligned with corporate ESG goals.

Your team, acting as RetailMax’s ML project group, must redesign the training pipeline to balance accuracy, cost, and sustainability—delivering a model that performs efficiently, supports business priorities, and reflects responsible AI practice.

Download the efficiency challenge template​

Use the link below to download RetailMax Efficiency Challenge Template. Work with your group to fill it out as you complete each step.

Optional: Download a copy of the workshop slides​

Activity instructions​

Work with your group to complete the following steps:

Review the context​

Start by reviewing RetailMax project scenario. Focus on the company’s goal, and the current challenge of balancing accuracy, cost, and sustainability.

Design your ML training workflow​

  • Sketch or outline your optimized ML pipeline.
  • Include key stages such as data ingestion, preprocessing, training, validation, and deployment.
  • Identify where you plan to introduce efficiency improvements.

Apply optimization strategies​

  • List 2–3 optimisation techniques your team would apply and explain briefly how each improves efficiency or sustainability.

Select your performance metric​

  • Choose one performance metric your team would use to track model training effectiveness (e.g., F1, recall, latency).
  • Explain how it supports RetailMax’s business goal.

Balance your trade-offs​

  • Describe one trade-off your team had to make between accuracy, cost, and sustainability.
  • Explain your reasoning and how you would justify it to leadership.

Regroup and share insight​

Return to the main session after 20 minutes to discuss key insights and recommendations.

Action item: Activity share out​

  • Which stakeholder’s priority did your solution favor most (operations, data science, finance, sustainability)?
  • What was your team’s most impactful efficiency decision?
  • What would you adjust in future iterations of this pipeline?