endorsedCompleted

AI-Powered Time-Series Forecasting for A.P. Moller - Maersk

Services Covered:

AI and Machine Learning • Custom Software Development • Legacy Application Modernization

Industry

Logistics & Transportation

Duration

5 months

Budget

50K - 199K

Start Date

1 January 2025

Team Size

1-5 Employees

Engagement Model

Offshore

Project Capability Score (PCS)

73/100

Good

Project Capability Score (PCS) estimates how capable Devox Software is of successfully delivering a project like this, based on its past experience, track record, and ability to handle similar work. Learn more about PCS

Client

A.P. Moller - Maersk

Copenhagen, Denmark

Project Summary

Devox Software designed and delivered a customized AI-powered time-series forecasting solution for A.P. Moller - Maersk, enhancing the client's Python neural network libraries and integrating additional external data sources. The work improved forecast accuracy by 24%, enabling better sales and inventory planning and reducing overstock and stockouts. The team delivered on time, provided documentation and knowledge-transfer, and communicated regularly via Microsoft Teams and email.

Key Challenges

Improve the accuracy of annual and seasonal sales forecasting using AI-powered models.
Enhance in-house Python neural network libraries to better handle retail-specific time-series data.

Project Deliverables

Extension and optimization of in-house Python neural network libraries for sales forecasting.
Integration of external data sources such as weather data, holiday markers, and promotional campaign indicators.
Development, testing, and benchmarking of machine learning models across multiple product categories and regions.
A fully documented codebase prepared for internal use and future scaling.
Knowledge transfer sessions and internal handovers to the client's team.

Project Solution

Devox Software designed and implemented a fully customized AI-powered time-series forecasting system. The engagement included enhancing the client's existing Python neural network libraries, integrating external data sources (weather, holiday markers, promotional indicators), and developing, testing, and benchmarking machine learning models across product categories and regions. The vendor delivered a documented codebase and conducted knowledge-transfer sessions to the internal team.

Project Outcome

24% improvement in forecast accuracy, leading to more reliable sales and inventory planning.
Enhanced stock optimization across regions and product categories, reducing overstock and stockouts.
Improved revenue projections and demand planning around seasonal peaks and promotions.
Increased confidence in decision-making from operations and finance due to more precise forecasting.

Platforms

AI/ML Platform

Tech Stack

Python

Client Endorsement

Overall Review Rating

Timeliness

Cost Rating

Willing to Refer

Quality of Deliverables

“Thanks to Devox Software, the client saw a 24% improvement in forecast accuracy, resulting in better sales and inventory planning. The team was highly responsive and proactive, regularly updating the client on progress and adapting quickly to feedback. Their work was timely and well-executed.”
Vincent Clerc
CEO
Date: 08/04/2025