Issue link: https://read.uberflip.com/i/1381595
3 2 PA G E N A M E V O I C E O F T H E PA R T N E R S Addressing the people, process and technology challenges Organisations are investing heavily in data science and artificial intelligence capabilities to scale these applications, and although this represents a significant asset, often the value realised from these endeavours is underwhelming. As a global firm, Deloitte recognises the huge impact that artificial intelligence can have on businesses and understands that successful adoption is often dependent on overcoming people, process and technology problems that invariably constrain technology adoption. While model development has traditionally been the domain of data scientists, maintaining, deploying, and monitoring model performance is not – it sits at the juncture of business, data science and traditional technology delivery. Scaling and uplifting data science and machine learning capability requires a robust and repeatable delivery methodology, which drives consistent communication of data science outcomes across the business. Machine learning operations to scale ModelOps is a rapidly emerging field at the intersection of artificial intelligence, DevOps, and Data Engineering, seeking to manage the complexity inherent within maintaining both the underlying software and code base of a model and also the data pipelines relied upon to produce results. Building and embedding ModelOps within the delivery of artificial intelligence technologies helps to accelerate delivery of accurate models, and facilitate continuous development, ultimately generating greater business value and relevance. An Australian telecommunications provider was struggling with a predominantly manual process for classifying and addressing customer complaints. Prior attempts to produce an artificial intelligence-based modelling solution had been unsuccessful because model deployment was often challenging and slow, tool selection wasn't aligned to business processes and performance was not continuously monitored, creating challenges in assessing if the model had degraded and required retraining. Through implementing a fit-for-purpose ModelOps Workbench powered by AWS, including SageMaker for Automated Feature Engineering and AutoML capabilities, Deloitte and the telecommunications provider were able to establish an artificial intelligence-powered model that auto-classified over two-thirds of customer complaints at greater than 85% accuracy, all delivered from concept to production in under four weeks. Combining people, process and technology capabilities together with the cloud services that enable repeatable and automated delivery, provides organisations with the ability to realise lasting business value from artificial intelligence as we move into an ever-intelligent future. Find out more about Deloitte and AWS > 3 2 B A C K T O C O N T E N T S >