Operationalizing Machine Learning - A Deeper Dive
with Shreya Shankar, Hamel Husain and Josh Wills
What to expect?
A group of Berkeley researchers recently surveyed the challenges and opportunities of operationalizing machine learning (MLOps) across many industries and domains. This survey is unique in that it captured many vital aspects of MLOps that are seldom discussed, such as: - Eschewing complexity for more pragmatic approaches. - Misguided statistics claim that most ML projects fail. - The importance of processes and organizational design, in addition to tools. - A nuanced debate of config-based, declarative tools (i.e., YAML) vis imperative ones. - and many other aspects!