Making the world's railways safer
Sperry Rail partnered with Steamhaus and AWS to accelerate and de-risk their Digital Transformation journey. Together we are building industry leading, Cloud Native products, unlocking the true value of their data, automating analysis, and ensuring safer more reliable railways.
- Rebuilt Elmer AI from PoC to scalable app using AWS serverless architecture.
- Migrated EC2 workloads to Amazon ECS with AWS Fargate, cutting monthly costs by ~35%.
- Used Amazon Athena and DynamoDB for fast, efficient data processing.
- Built Zula data lake to unify and access historical rail data.
- 3-year project included AWS Well-Architected review and full platform modernisation.
Unlocking scalable AI for smarter rail testing
Sperry Rail Systems has been a leader in non-destructive rail testing by innovating new technologies and processes since their founding in 1928 by Elmer Sperry. In that time, the company has inspected more than 12 million miles of track and found more than 6 million defects. Their culture of innovation saw the company transform to a technology-first organisation that happens to use their technology for rail testing. While working on a proof of concept, the company found that innovation can come with problems of its own.
Using primarily ultrasound, the company inspects rails for track owners around the globe; every night, their tests involve 100 or more trains, each running over 10–100 miles of track. Analysts interpret the data to detect cracks or other issues, alerting the operator so that maintenance can be performed. The company developed an artificial intelligence (AI) tool to improve the speed and accuracy of their analysis. The only problem was that it worked so well that everyone wanted to use it, resulting in wait times of up to 12 hours to process data.
Sperry’s innovation team began developing the AI—named Elmer after company founder Elmer Sperry—as a proof of concept (PoC). The PoC was built on Amazon Web Services (AWS) to see if it could work, and it did. It helped analysts pick out trouble spots much more quickly and more accurately. As a result, Sperry saw high demand for access to the PoC. But Elmer couldn’t scale and throughput was limited, which led to long queues. Sperry turned to Steamhaus, because of our AWS expertise, for help with developing this PoC into a full-fledged application at scale.
From PoC to production: Utilising AI for rail efficiency
The project began after Sperry asked AWS how to develop the Elmer PoC into a scalable application and was referred to Steamhaus, an AWS Partner. The first step was to conduct a Well-Architected review, which helped determine Elmer’s strengths and opportunities for improvement. After the review, Sperry engaged Steamhaus to help rebuild Elmer from an effective PoC into an enterprise-grade solution.
Steamhaus and Sperry undertook a three-year collaboration to improve Elmer. Steamhaus identified two good starting points: move to serverless services and take advantage of automation features in AWS. Sperry had been using AWS primarily as a compute service, which wasn’t cost-effective at scale. So Steamhaus did more than simply rebuild the PoC. We helped Sperry to completely revamp the business, by changing how they worked and by inspiring new ways of thinking about how to use AWS services to set up opportunities for further advancement.
Through collaborative working, Steamhaus gained a deep understanding of Sperry’s ambition to develop as a technology company. And Sperry benefitted from the AWS expertise and modernisation framework that Steamhaus was able to bring to the project. Working together, we succeeded at designing, building, deploying, and scaling a full blown, production-ready Elmer. During the process, Sperry was also able to take what they learnt to expand their vision for the company’s future.
Driving faster, more cost effective railway analysis with Elmer
The key to the engagement between Sperry and Steamhaus was not being confined to the goal of rebuilding the Elmer PoC. The in-depth discussions between the companies focussed on the outcome, rather than on the initial PoC code. It was a complete reimagining of what was possible, from a promising PoC to a serverless, web-based railway-analysis solution.
This still had to consider business realities and use Sperry’s resources efficiently. Steamhaus applied our AWS knowledge to get the best technical solution while minimising costs. For example, we containerised the company’s Amazon Elastic Compute Cloud (Amazon EC2) workloads using Amazon Elastic Container Service (Amazon ECS). This solution uses AWS Fargate, a serverless, pay-as-you-go compute engine. By paying for only what they use, Sperry reduced the monthly cost of Elmer by about 35 percent.
Sperry uses Amazon Athena and Amazon DynamoDB to perform at speed. Elmer needed data to train the model, so Sperry’s next step was to build a data lake that made all previous data accessible to users and to Elmer. That data lake—called Zula, after Elmer Sperry’s wife—means that analysts can access previous testing data to better understand the results of the current scans.
Working together, Sperry and Steamhaus were able to transform Elmer from an experiment to an industry-leading game changer.
Headquartered in the US, Sperry Rail Systems has been a leader in non-destructive testing since their founding by Elmer Sperry in 1928. The company’s 500+ employees work on railways around the world— from heavy haul networks to metropolitan systems—and have inspected more than 12 million miles of track and found more than 6 million rail defects. The company continually innovated new technologies and processes, until they saw a future where that capacity to innovate could transform them into a technology business.Location
Shelton, USAIndustry
Transportation
Services
AWS PlatformsMigrate & ModerniseShare
Shaping the future: ongoing modernisation and growth
Sperry is committed to maintaining momentum, continuing the modernisation programme, and developing further improvements through their ongoing collaboration with Steamhaus. They are also focussed on team enablement and upskilling their teams, including creating an internal Cloud-Native Centre of Excellence. Sperry aims to ensure that the business is fully equipped to run new products and can continuously iterate, in accordance with best practice and enterprise-scale standards for security, resilience, and cost-efficiency.
We are also working with the wider Sperry US team to apply the lessons learnt from the Elmer and Zula work to other business-critical systems through a second MAP engagement.
A key driver of Elmer’s success was the reimagining of Sperry’s business. The company does not see themselves in the railway service business—they now see themselves as a data- and AI-driven software company. Steamhaus and Sperry are on track to continue to innovate, both on solutions and on the scope of the business.
