← sources

Transitioning away from DE

source: reddit (reddit:dataengineering)published 2026-10-02 23:14ingested 2026-10-03 02:36

retained text

Has anyone thought of transitioning out from DE due to AI?  
All I do everyday is just prompt and scroll till copilot generates code.   
Building a semantic layer isn’t exciting personally as I don’t enjoy the business aspect of it as much and think of it as more of a data labeling and analyst problem than an engineering problem (which I am interested in)  
Also, There is a fundamental problem with “I am building a semantic layer” and marketing that as a skill as it is dependent on how much context you have of the business. The less tenure you have spent in a company, the less you know about the business which makes it harder as a transferable skill imo.

My understanding is that working on building trustworthy AI outputs by using a feedback loop is an engineering problem to solve. Which is why I feel going down the observability path is a good idea.   
I heard these opinions on observability from AI leaders at conferences too so there might be a bias.  
Thoughts from fellow DE’s looking to transition out? (Or from one’s who want to continue and why)

document fetched 2026-10-03 00:24 · canonical https://reddit.com/r/dataengineering/comments/1ww83g5/transitioning_away_from_de