MLOps with Databricks // Maria Vechtomova // #314

·53 min·5 clips

Maria argues that notebooks harm the MLOps lifecycle more than any other tool.

The episode begins with host Dimitrios introducing Maria Vechtomova, an MLOps tech lead at Aho Del Geise and co-founder of Marvelous MLOps. Maria shares her background in MLOps, spanning over eight years, including building custom tooling before it became mainstream. She explains her motivation for starting to post on LinkedIn and write articles to counter misinformation in the MLOps space. The conversation then focuses on Databricks, with Maria discussing why she chose it as a platform, referencing a survey by the Ethical Institute of AI that ranked it as the top tool. Maria criticizes the overreliance on notebooks in Databricks training, arguing they hinder production readiness. She details the pros of Databricks, such as Asset Bundles for development and deployment, and the cons, including limitations in the feature store and development processes. Maria talks about when not to use Databricks, such as for low-latency serving or high request rates. She also touches on MLflow's recent updates, including LLM features and serving capabilities. The discussion covers practical use cases, like recommender systems, and how Maria's team integrates Databricks with other tools like Azure Functions and Kubernetes. Maria shares insights on data quality challenges when consuming data from other teams and attempts at data contracts. Towards the end, she promotes her cohort-based course on MLOps with Databricks, starting in May, and her upcoming book with O'Reilly, set for release next year. The episode concludes with a discount code for the MLOps community.

As heard by us

A grounded Databricks review where managed convenience meets the stubborn realities of MLOps.

Databricks comes through as a practical case study in the tradeoffs behind managed MLOps. Maria, an MLOps tech lead at Aho Del Geise and co-founder of Marvelous MLOps, maps where the platform smooths the path and where its built-in assumptions can start to pinch.

Read the full review in PlayNext →

Why you'd press play

You want Databricks tradeoffs spelled out by Maria, an MLOps tech lead and Marvelous MLOps co-founder who teaches machine learning and MLOps.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from MLOps.community