Using Databrick’s MLFlow to train, track, reuse and deploy their models
We were delighted to be joined by Dr. Stefano Bosisio, Machine Learning Engineer at Trustpilot, who presented his views on the benefits that can be gained by Data Scientists if they utilise Databrick’s MLFlow to train, track, reuse and deploy their models.Watch the full session.
MLFlow In Recap
Our three takeaways from the discussion with Stefano were:
- The importance of a “log book” for Data Scientists – detailing every iteration, change, adaption and evolution of experiments – ensuring a structured, cohesive recording of each stage in the process.
- MLFlow provides this service with a clear, sharp, intuitive graphical interface – with a load of interoperability behind!
- MLflow is open source and the main features come for free, although some more advanced deployment products have a cost to consider.