Can AI automate MLOps enough for data scientists to avoid it?
I come from a strong math/stats background and really enjoy the modeling, analysis, and problem-framing side of data science (e.g. feature engineering, experimentation, interpreting results).
What I’m less interested in is the MLOps side — things like deployment, CI/CD pipelines, Docker, monitoring, infra, etc.
With how fast AI tools are improving (e.g. code generation, AutoML, deployment assistants), I’m wondering:
Can AI realistically automate a large part of MLOps workflows in the near future?
Are we reaching a point where a data scientist can mostly focus on modeling + insights, while AI handles the engineering-heavy parts?
Or is MLOps still fundamentally something you need solid understanding of, regardless of AI?
For those working in industry:
How much of your MLOps work is already being assisted or replaced by AI tools?
Do you see this trend continuing to the point where math/stats skillsets become more valued by employers?