This past week, Microsoft announced the acquisition of Osmos: a Seattle-based data engineering startup focusing on making the frontier “AI Data Engineer” the next hire on your team. Will Osmos be able to create your point-to-point autonomous ETL with production ready PySpark Notebooks given on the source and destination schemas? Create your Lakehouses and OneLake datasets? That’s the idea.
Osmos has been in the Fabric toolkit as a Workload for some time, but appears to have been quietly deprecated in the last few months/weeks. Admittedly, I didn’t have much exposure to the workload. However, the announcement this week set my timeline on fire with excitement at its potential.
Empowering engineers with the ability to hand wrangling and notebook configuration off to an agent with conversational instructions. Sign me (and every manager, everywhere) up.
How productive though, is Osmos this for mature data engineering teams with strict standards, established OneLakes and envrionment libraries, and residency restrictions? Likely not AS productive as a team just getting started, but the tool seems targeted more at those groups anyway. Lower the barriers of entry to getting started in a best practice way. Which really is the mantra Fabric is trying to live by anyway.
-DS
Full Data Extract 
Leave a comment