
Differences Between Shipping Classic Software and Operating ML Models with a Lead MLOps Engineer at TMNL Simon Stiebellehner, and neptune.ai CEO Piotr Niedzwiedz
ML Platform Podcast
00:00
Model Registry
The way we use it currently, is really focused on that model artifact. It gives us the links to the pipeline that produced it. So basically it's one of the outputs of that pipeline. And what we try to do is we link back every single artifact produced to that specific pipeline and that pipeline holds all the information about what jobs were in there,. What images were used in that pipeline and so on. We can basically suck it all together into one big picture. But personally I would be a lot more happier if it was a pipeline registry, giving a nice overview of pipelines.
Play episode from 50:23
Transcript


