Currently, working on understanding the inefficiences in the way we train, retrain and manage our machine learning / deep learning models. Also improving our practises on how we manage data, model and experiment versions knit together with the likes of MLFlow and other tooling.
Links This has led to creation of an open source project called sentinels
Currently, architecting release process engineering for the development team at omnius. Our customers are either enterprise on-premise customers or enterprise cloud customers. In both the scenarios, we have restricted or sometimes no access. We are a team 30 engineers, data scientists and devops. Because of the sensitivity of the data that we work on, we decided to run our own mini data center from our HQ in Berlin. As part of release process engineering, we are creating various environments running kubernetes to support our branching strategy in order to provide good, tested software to our customers.