S3 | Ep 38 | Building Value with MLOps with Sami Rahman, Head of Data Engineering and Data Platform at Penguin Random House
In Episode 38 of Season 3, of Driven by Data: The Podcast, Kyle Winterbottom is joined by Sami Rahman, Head of Data Engineering and Data Platform at Penguin Random House, where they discuss how to build an MLOps capability that drives business benefit, which includes;His journey from Business Psychologist to MI5 reject to Data Leader Being told to give up because he would never work in data without a computer science degree Being unemployed for 1 year and receiving 326 rejections Building an entire MLOps system for £50,000Reducing the time to build models from months to days Reducing the time to productionise ML models from 5 months to minutes sGenerating a 200% increase in performance across all 12 measures of success Why MLOps forces you to think about how the model will drive valueHow MLOps can be beneficial across the whole D&A value chain The differences between MLOps and AIOps Why MLOps isn’t new DevOpsThe future of MLOps and the relationship with Genertaive AIChoosing the right MLOps tech and tooling Why you need to explain MLOps in less than 2 sentencesWhen you should start your MLOps journey The usual pitfalls both technical and business that people should be aware ofThe ethical considerationsAI Explainability - how to understand the outcomes in a business context Why we should build an “evil” AI to tackle bias The skills you need to execute successful MLOps