ACL Digital

Home / CaseStudy / ACL Digital Transformed Data Science Pipelines for a Global Rail Transport Leader

ACL Digital Transformed Data Science Pipelines for a Global Rail Transport Leader

ACL Digital Transformed Data Science Pipelines for a Global Rail Transport Leader

ACL Digital Transformed Data Science Pipelines Banner

Overview

A global leader in rail transport sought to revolutionize its data analytics processes by industrializing its Data Science Pipeline. Operating across passenger transportation, signaling, and locomotive sectors, the company aimed to enhance odometry and radioscopy mobility analytics efficiency to stay competitive in a rapidly evolving industry.

Download Case Study




    Challenges

    Extracting, managing, and operationalizing large volumes of odometry data

    Building robust pipelines to train, deploy, and automate anomaly detection in mobility analytics

    Ensuring seamless deployment of machine learning (ML) models and their integration with user interfaces for performance monitoring

    Minimizing manual intervention to reduce operational inefficiencies and errors

    Solutions

    ACL Digital’s expertise in data engineering and machine learning implementation was instrumental in transforming the company’s data science processes, delivering measurable operational efficiency and accuracy improvements.
    • Extracted code from Jupyter notebooks and created a data pipeline to transfer odometry data to a Minio bucket efficiently.
    • Designed a training pipeline for anomaly detection and classification of anomalous events.
    • Deployed ML models as OpenFaaS functions and implemented a CI pipeline to automate model deployment.
    • Developed and tested a user interface application to display the performance of the deployed models.

    Outcomes

    Transformed Data Science Pipelines
    Automation

    Enabled the automated detection and classification of anomalies, significantly reducing manual workload and errors

    This marked a substantial improvement over traditional OCR methods, which struggled with the degraded quality of the images

    Provided a robust infrastructure for deploying future analytics use cases, enhancing adaptability to evolving business needs

    Transformed Data Science Pipelines
    Scroll to Top