Back to customer stories

DevOps for Industrial Artificial Intelligence

Increare in conversions
Increare in conversions
Increare in conversions

Sentian needed a solution that would help them scale, organize and automate all the components they had on their cloud solution. So, Microtica provided them with the ability to create blueprints and adapt for each client.

Before Microtica: failing to achieve scalability for 20-30 clients at the same time by sticking to traditional CI/CD

Due to extensive research programs in applied AI, Sentian has the ability to deliver leading-edge products that give its customers a competitive edge.

The company offers three Industrial Artificial Intelligence products: ICON (Intelligent Control Optimization) that boosts existing control and automation solutions with AI, predictive maintenance for large scale implementations in distributed environments, and its world-leading nonlinear optimization solver for production planning, supply chain optimization, and personnel planning optimization.

Since the beginning (2016), has been helping many organizations from the energy, manufacturing, and process industries to optimize their production processes, and has been recognized for its leading AI technology and solutions multiple times.

Sentian has developed an Industrial Artificial Intelligence solution that optimizes the industrial production processes through the application of AI engines for intelligent control optimization.

In the beginning, their solution was hosted either on-premise or as a single instance on the cloud. They needed a solution that would help them scale, organize, and automate all the components they had on their solution in the cloud. These components are covering different aspects of the solution, from a controller that directs the data you put in, a planner that executes the actual planning, to components that train and execute AI models.

The team of Sentian needed a different configuration of the modules for each specific case they are working on. All of these modules had to together be scalable and moved to the cloud. The solution that they wanted to move to the cloud had to be sustainable and replicable for each new client. With traditional CI/CD, they couldn’t achieve the scalability they need for 20-30 clients at the same time.

Coming across a solution: setting up the entire solution and infrastructure for machine learning and AI on the cloud

As Sentian works with a large number of companies from many different industries, they all have different needs and preferences. This means that for each new client, they had to re-configure the solution, which was too complex and time-consuming.

They evaluated different CI/CD solutions that would help them to set up and operate the solution in the cloud.

They had several main criteria while choosing the tools to use:

  • an ability to build and automate the entire infrastructure setup in the cloud,
  • an ability to automate the delivery of their modules on Kubernetes,
  • a solution that is enough flexible and extendable to support their self-service model,
  • and the most important criterion was the ability to replicate the whole setup for
  • different projects in a matter of hours.

None of the evaluated tools gave them the ability to scale the solution the way it would support their business model. Then, they started evaluating Microtica:

  • Predictive maintenance and anomaly detection
  • Intelligent automation and control
  • Intelligent optimization

Sentian started with a PoC (Proof of Concept). The initial goal was to prove that the entire solution and infrastructure for machine learning and AI can be scalable and set up on the cloud. Moreover, we provided them with a solution that will organize all their modules and allow them to onboard new customers much faster.

The architecture they set up is based on two parts:

  1. The infrastructure, in which they set up the networking, Kubernetes, databases, storages, etc. This is the part that supports and runs Sentian’s solution.
  2. The Kubernetes integration, which automates the infrastructure release of Sentian’s components. In Microtica, they basically created a catalog of Sentian’s components, allowing them to build a configured system by taking the needed modules for the specific use case from here.

After proving this, the next phase involves the replication of the solution for real projects.

Outcome: discovering the power of Microtica

Microtica was able to cover all of the criteria Sentian expected from a tool that handles crucial automation processes within their solution.

Microtica provided ready-to-use components they could use immediately. Microtica’s components represent reusable pieces of code that can handle the cloud infrastructure without needing to rewrite or adapt the infrastructure every time you want to use it in another project or product you are developing. This way, Sentian’s team was able to start building their infrastructure without having to worry about the complexity involved in developing, testing, and provisioning the underlying infrastructure in the cloud.

The Kubernetes integration gave them the ability to immediately start shipping the modules. No additional scripts or complex integration had to be made.

One of the biggest benefits of Microtica was that after adding all the modules into Microtica and automating them, the Sentian team was able to reuse the created modules and build new instances from their solution for different projects within hours.

Ultimately, they will be able to create blueprints for different types of clients and adapt their solution for each new client by only having to make minor or no changes.

This means that instead of data scientists having to spend two or three weeks adapting the entire pipeline to the new project, they will be able to set the entire infrastructure up in a few hours.

Microtica runs the automatization of the entire Sentian solution. Now, Sentian can onboard new customers in a day or two.

Technologies: AWS, Python, NodeJS, Kubernetes, Docker, gRPC, Ambassador Edge Stack, Auth0

Less cost on third-party tools
80 hours/month
Saved on platform management
Faster customer onboarding

You may also like

See all customer stories