I have used MetaFlow to orchestrate some batch scoring previously and really like it. Recently I have done a lot of experimentation - painfully writing my own code. I am considering MLFlow to become more efficient. I guess MetaFlow could also be used but I think MLFlow is more suitable for this use case OOTB. Could some experts please provide some feedback regarding this? Thanks.

@happy-wolf-7852 can you elaborate on your usecase?

So it is about taking snapshots/artefacts of training data, logging experiment settings like hyper parameters and results - e.g. lots of metrics, plus the used source code (push/commit). My current understanding is, that all this is quite simple using MLFlow OOTB (locally up to having a server and dedicated database). Maybe there is an example project to do similar with MetaFlow without having to reimplement the wheel? Hope this makes sense?

all of this information automatically gets logged by metaflow for every execution

ok thanks - I will revisit this aspect (plus new tiles capabilities) as I mainly focused on batch processing use thus far.