During my career as an evolutionary biologist, I spent years studying how life adapts to its environment. I examined different angles and domains in biology and it forged the way I think about everything today. Then I pivoted into data science, then data engineering and infrastructure. And somewhere between writing Terraform modules and debugging silent pipeline failures, I noticed something: the infrastructure I was building felt static. For example, even after setting up alerts and monitoring systems, we still had to migrate, fix bugs, and upgrade ourselves. I wonder now if there is a less rigid yes robust way to build data infrastructure.

I believe that biological systems have been solving the same problems for a very long time: they evolved to be resilient, redundant when necessary, maybe anticipate, and self-repair themselves. Can we actually leverage our current understanding of biological systems to make data infrastructures more “alive” and efficient?

What’s Next?

I am setting an experimental and thought process project here grounded in use cases found in DevOps and MLOps.

Go to this repo to follow the implementation of these different concepts.

→ living-infrastructure on GitHub

Isabelle Vea.
Chicago, August 2026.