In the midst of a global pandemic, people and organizations have changed their behaviors, ranging from spending to transportation to business operations. These changes get reflected in the dataset shift and may undermine many model assumptions, thereby impacting the performance of ML applications. In this blog post, data scientists from Quantum Black recommend steps to adapt ML pipelines to a post-COVID-19 world, including by strengthening and recalibrating systems for model management and governance; leveraging new data sources or increasing the frequency and granularity of internal data collection; and using new modeling approaches (including more techniques that are more interpretable, facilitate uncertainty quantification, and/or adapt to volatile conditions).