MLOps in theory and practice

Recently first episode Podcast by Tinkoff Education and the Higher School of Economics. It was about MLOps. Now AI/ML topics are on everyone’s radar: everyone is trying to use LLM to optimize existing processes, someone is trying to come up with breakthrough ideas, and someone is afraid of these changes. Universities now have a large number of data science and machine learning courses, but real companies often need not only to create a good model, but also to make a bunch of other squats, which can be called MLOps. To describe the essence, you need a clear connection between Data engineering, App engineering and, in fact, ML engineering. Moreover, we need built-up processes of working with data as prequisites for effective work on ML models, and App engineering is needed so that trained models work well in production and serve queries.

Participated in the discussion Dmitry Ushanov, Technical Director of ML Platform

• Evgeny Sokolov - Academic Head of PMI FKN HSE
• and

By the way, you can find out about new episodes of this podcast and other educational news from Tinkoff on the Tinkoff Education telegram channel. (@tinkoff_fintech).

P.S. I've already read two posts on this channel. (1 and 2) That little whitepaper about Google's MLOps that we used as a podcast discussion framework.

#ML #Devops #Data #AI #Software #Architecture #Processes