Data Reply and Seldon hosted a joint event, the Data Mash Meet Up, in partnership at the Reply offices in Munich on 16th November to discuss Predictive Maintenance using MLOps.
Data Reply is the Reply Company specializing in helping customers to become Data-Driven by providing state of the art solutions leveraging Distributed & Cloud technologies. They provide architecture and development solutions for Big Data applications, services, and infrastructure such as Data Platforms, Real-Time Decision Engines, and Machine Learning Applications. We are proud to be partnering with them and to have had the opportunity to come together this Autumn in Germany to learn about scalable solutions and best practices for Predictive Maintenance leveraging advanced MLOps capabilities from both Data Reply and Seldon.
Predictive Maintenance- big picture and its use-cases
The first of the three talks that were featured in this meetup was by ML Engineer at Data Reply, Peerapon Wechsuwanmanee “Predictive Maintenance- big picture and its use-cases”.
Data Reply covered some key considerations for companies considering Predictive Maintenance including:
- The market impact
- Predicted to grow from $4.2BN in 2021 to reach $15.9BN by 2026
- Industry use-cases
- Common challenges in deploying, monitoring & scaling MLOps.
How to overcome challenges in continuous development, monitoring, and scaling in MLOps
He described an MLOps blueprint for predictive maintenance, which included best practices, a typical workflow, and a reference architecture. The technologies presented represented an end-to-end system, including Seldon as the main serving and monitoring component of the stack.
Next up was ML Engineer at Data Reply, Marwan Fahmi, who gave a talk titled, “How to overcome challenges in continuous development, monitoring, and scaling in MLOps”.
Hands-on MLOps workflow with Seldon
Our Solutions Engineer, Andrew Wilson, presented a talk titled, “Hands-on MLOps workflow with Seldon”.
Andrew presented an overview of Seldon technologies, including open source tools like MLServer, Seldon Core, and Alibi Explain/Alibi Detect, as well as our enterprise product, Seldon Deploy Advanced. Seldon then performed a live demo to show how to train two image classification models to detect defects in steel strips, deploy these as an A/B test, add model metadata, and create a drift detector.
The talks highlighted the importance of why organisations should start looking at Predictive Maintenance now and become data-driven, leveraging Data Reply’s MLOps capabilities with Seldon’s market-leading Machine Learning technology.
Following the talks, attendees from academia and industry enjoyed drinking German beer and chatting about their MLOps backgrounds, challenges, and solutions. We believe it is just as important to host our own MLOps London meet ups as well as visit other MLOps related meet ups such as the Munich Data Mash that is hosted by Data Reply.
The intention is to continuously highlight the innovative MLOps industry, learn and bounce ideas off each other, and network with like minded people. We’re happy to say we achieved all three of these goals during the Munich Data Mash event. See you at the next meet up!
Tom joined Seldon as VP of Partnerships after gaining more than 10 years of experience in the industry at a variety of Technology organisations. Prior to Seldon, Tom looked after strategic alliances for technology intelligence platform, Snow Software, and before that, he spent more than eight years in enterprise software sales, working at companies like Adobe and Oracle. He started his career on the IT services side at two of the world’s largest Global Systems Integrators, HCL Technologies & DXC Technology. Having had first-hand experience with these types of organisations prior to joining a vendor, he’s gained a strategic insight into what potential partners would be looking for in a software provider, and has brought this knowledge to Seldon.