Cognizant MLOps Model Lifecycle Orchestrator Speeds Deployment of Machine Learning Models from Weeks to Hours Using AWS Solutions
Organizations across the globe increasingly recognize that artificial intelligence (AI) and machine learning (ML) are critical tools for business innovation, but using these tools at scale can be challenging. Organizations need a robust ML operations (MLOps) framework to scale quickly and be successful at getting models into production while maintaining, monitoring, and continuously improving at the right speed and agility. To automate and reduce the time and effort its clients need to deploy and maintain ML models, global systems integrator and Amazon Web Services (AWS) Partner Cognizant created an MLOps solution that supports organizations’ ML modernization needs.
In collaboration with the AWS Partner Solutions Architect and AWS Solutions Library teams, Cognizant built its MLOps Model Lifecycle Orchestrator solution on top of the MLOps Workload Orchestrator, an extendable framework that provides a standard interface for managing ML pipelines. By applying its intellectual property and best-practice AI product build capabilities, Cognizant built a solution that reduced the deployment times for new ML models from weeks to hours.
Using the MLOps Workload Orchestrator from the AWS Solutions Library was a huge benefit in terms of resource costs and time savings.”
Adapting to Changing Needs
Cognizant supports some of the leading global enterprise organizations across industries. The company’s consultative approach helps transform its clients’ business, operating, and technology models. In particular, Cognizant’s data and analytics team helps organizations accelerate business growth using AI and ML.
Cognizant recognized early on that clients needed to invest more in AI and ML technology to make better business decisions. It identified that there was a gap in how clients were able to realize an efficient MLOps solution that could promote collaboration between data scientists and IT teams to make building, testing, deploying, and maintaining AI and ML models easier. Because the data used in these models can change rapidly, along with consumer behavior, companies need agility to quickly adapt their models. An MLOps solution helps businesses manage the complexity of maintaining ML models and make business decisions based on the results of those models. Cognizant considered creating its own solution from scratch but chose to build its solution on top of the MLOps Workload Orchestrator from the AWS Solutions Library to save time and resources in development in addition to using AWS services that the company and its clients were already familiar with.
As Cognizant developed its solution, it collaborated closely with the AWS Partner Solutions Architect and AWS Solutions Library teams on the market requirements and the features and functionality required to support its clients’ short- and long-term needs. AWS also added features needed for Cognizant’s solution to the road map for the MLOps Workload Orchestrator. “We have deep technical and business collaboration with the AWS team, which means that we are always superfocused on business outcomes and the technology that will deliver client success,” says Nejde Manuelian, head of data and AI global partner solutions at Cognizant.
Creating a Fast and Accurate MLOps Solution
By building its MLOps solution on top of the MLOps Workload Orchestrator, Cognizant created a solution that works seamlessly alongside other AWS services that clients use. For example, Cognizant’s solution uses Amazon API Gateway—a fully managed service for developers to create, publish, maintain, monitor, and secure APIs at virtually any scale—for inferencing so that the company can build its own accessible user interface. The solution also uses Amazon SageMaker, a fully managed service to build, train, and deploy ML models for virtually any use case with fully managed infrastructure, tools, and workflows. The result is a user-friendly solution that benefits from the ongoing innovation of AWS services. “Our goal was to provide a simple user interface for users to interact with and an interface for MLOps engineers to deploy,” says Ajay Sharma, chief architect at Cognizant. “For each model, users can input the parameters needed into the user interface and deploy the model to the target infrastructure.”
Cognizant has seen significant cost savings by building its solution using the MLOps Workload Orchestrator from the AWS Solutions Library instead of developing a solution from scratch. “To build a comprehensive solution on your own could take years and millions of dollars,” says Sharma. Additionally, the company would have needed to invest personnel in ongoing maintenance and development of an internal solution. By using AWS services, Cognizant was instead able to develop a solution in 3 months and help its clients realize benefits right away. “Using the MLOps Workload Orchestrator from the AWS Solutions Library was a huge benefit in terms of resource costs and time savings,” says Sharma. “Once our MLOps Model Lifecycle Orchestrator solution was in place, we could reduce the timeline for deploying every new ML model from weeks to hours.” Cognizant follows a defined, standardized process and uses automation for ML model building and deployment, resulting in 80 percent cost savings in time and effort.
Cognizant’s MLOps Model Lifecycle Orchestrator solution helps the company’s clients improve the accuracy and speed of their ML models to drive business forward. Through this solution, Cognizant has defined and configured custom model monitoring metrics that calculate data drift and model health metrics, achieving over 95 percent model accuracy over time. The solution has also improved model runtime by 50 percent, supporting the agility and speed that Cognizant’s clients need. Overall, Cognizant’s solution acts as a business driver that supports organizations as they optimize the way that they run their businesses. For example, one of Cognizant’s clients uses the solution to deploy models that support critical business decisions, such as rapidly identifying the changes needed to products and services in order to adapt to cultural differences across different European countries during international expansion.
Staying on the Cutting Edge
Because MLOps is a new and evolving field, Cognizant plans to continue developing its solution as AWS adds capabilities to its services. “Our MLOps Model Lifecycle Orchestrator solution delivers better business decisions,” says Manuelian. “Our goal is to work with partners, such as AWS, to help engineer our clients’ businesses so that they can anticipate customer needs and act to meet them with the speed and insight of human intuition.”
Global systems integrator Cognizant transforms its clients’ business, operating, and technology models for the digital era. Its data and analytics team helps enterprise organizations accelerate business growth using machine learning and artificial intelligence.
Benefits of AWS
- Reduced deployment time of machine learning models from weeks to hours
- Achieved 80% cost savings due to standardized processes and automation
- Improved model runtime by 50%
- Achieved over 95% model accuracy over time using model monitoring metrics
AWS Services Used
MLOps Workload Orchestrator
The MLOps Workload Orchestrator solution helps you streamline and enforce architecture best practices for machine learning (ML) model productionization.
Amazon API Gateway
Amazon API Gateway is a fully managed service that makes it easy for developers to create, publish, maintain, monitor, and secure APIs at any scale.
Build, train, and deploy machine learning (ML) models for any use case with fully managed infrastructure, tools, and workflows.
Organizations of all sizes across all industries are transforming their businesses and delivering on their missions every day using AWS. Contact our experts and start your own AWS journey today.