Business Impact

  • 30%

    Reduction in Forecasting Errors

  • Increased Process Automation and Enabled Better Management and Visibility

  • Scalable, Robust and Flexible Solution Reduced Machine Downtime

  • 100+

    Accurate Failure Predictions for Machines

Customer Key Facts

  • Country : India
  • Industry : Manufacturing

Problem Context

The client is a recognized leader in the manufacturing of diesel engines, agricultural pump sets, and generating sets providing multinational services with their state-of-the-art manufacturing units in India. They have approximately 590+ machines/gensets of two types (CRDi and Power Car) in various locations.

IoT devices deliver sensor data to the Remote Monitoring application every 30 seconds, and the data is saved in a SQL database. Any failure is reported by the client, and it is noted in CRM software for maintenance.


  • Custom Models had to be developed for each of the 50 failure types to achieve better accuracies with limited failure data points.
  • The behavior of some of the failures was unexplainable by the sensor data.

Technologies Used

Amazon S3
AWS Lambda
Amazon SageMaker
Amazon Cloudwatch
Amazon Quicksight
Amazon SNS
Amazon Glue
Amazon DMS


Quantiphi evaluated and comprehended the client’s IoT data throughout the evaluation phase in order to identify suitable approaches to deal with the data set.

Quantiphi developed an ML Predictive Analytics solution to forecast non-functional problems at least 10 hours in advance and notify relevant stakeholders to execute essential maintenance before they occur. A retraining pipeline that uses recent data to improve the performance of ML models was also included in the solution.


  • Increased automation and better management of requirements to facilitate quick equipment maintenance and plan for service requirements in advance.
  • Scalable, compatible, and flexible solutions and infrastructure enabled failure predictions for hundreds of machines.

Looking for similar project?

Let's Talk

Get your digital transformation started

Let's Talk