From Silos to Synergy: Unifying DevOps, DataOps and ModelOps for AI-Powered Operations

For Business and IT to collaborate, for a Development and Operations team to effectively forge together for efficient operations, we have heard a lot about evolution of “DevOps” -> "DevSecOps" -> "BizDevSecOps" and so on. You can read more about this in my article -> “How BizDevSecOps drives value & Reshapes Organizational Structures?”

Tremendous amount of data is generated in last few years. This requires an approach to data analytics and insights that emphasizes communication, collaboration, integration, automation and data management and data consumption across an organization. This requires the need for “DataOps” which addresses the challenges related to data quality, integration and governance.

With hype in adoption of Artificial Intelligence in day to day application and business operations, solutions around “AIOps” started trending with focus on improving efficiency, agility and reliability of IT Operations. “MLOps” also came into play with focus on operationalization of Machine Learning Models. MLOps aimed at streamlining and optimizing the lifecycle of Machine Learning models which are deployed in an organization. While MLOps focuses only on particular Model Optimization, “ModelOps” focuses at an enterprise level of an organization to implement AI solutions, ?end-to-end governance and life cycle management of advance analytics, AI and decision models. MLOps is a subset of ModelOps.

Any CIO trying to drive the adoption and industrialization of AI across an organization should create a strategy to include ModelOps in an operating model. For efficient and AI enabled Next Gen Intelligent application operations, “DataOps”, “DevOps” & “ModelOps” needs to be combined under the roof of AI Center of Excellence.

ModelOps serves as a streamlined and transparent approach for operationalizing AI within an enterprise. It ensures management of business accountability to drive impactful business outcomes, by creating a robust process involving various stakeholders (ex: Finance, Procurement, Logistics, etc).

In addition to DevOps roles, additional roles such as Data Engineers, Data Scientist, AI Engineers would work cohesively as part of one team to achieve common goal to deliver Operational Excellence. Lets take an example of scenario which describes a real world use case combining DataOps, ModelOps & DevOps.

Case in a point

A Telecommunication equipment manufacturing company uses SAP ERP for inventory management, service provisioning and customer billing. Company is exploring an option to minimize downtime by implementing predictive maintenance, billing leakage, enhance end user experience and so on. Lets see how AI Center of Excellence combining DataOps, ModelOps and DevOps can help customer achieve this.

DataOps

• DataOps team will work on extracting / collecting & Analyzing relevant data and statistics across – Past Maintenance Orders, Monthly billing statistics, Customer Churn, Payment failures, Abandon cart, etc.
• Roles involved would be Data Engineers, Data Scientists who will work closely with Business value aligned “DevOps” teams (ex: Customer Billing & Interaction, Mfg. Operations, etc)

ModelOps

• ModelOps team will work closely with various business stakeholders – Finance, Customer Service, Mfg. and perform research on how AI can be leveraged to solve business problems.
• They will also closely work with DataOps team to develop and train ML models which focuses on predictive maintenance, billing leakage, analyze customer churns, customer buying patterns, etc.
• Once these models are finalized – deployment, further maintenance & enhancement (training with new data, enhance algorithm, etc.) will be the responsibility of this team.

DevOps

• DevOps team will ensure that all the required changes to the relevant applications in SAP are done
• Teams are organized based on Business Value Streams (ex: Customer Billing & Interaction, Mfg. Operations, etc.)
• Automation is the key focus – CI / CD pipeline

For more use cases refer my articles – “Empowering the Intelligent Enterprise: Harnessing the power of AI in Revolutionizing Business Ops” & “Maximizing Business Value: The Evolutionary Path of AI from Descriptive to Prescriptive !”

Any feedback, pls provide your comments / suggestions -- Mehul Chopra

***Pls note**** : These views are my own and not the views of my employer