A few words about me.

I started in ML, training computer vision models for medical imaging. After a year I came to think the most interesting AI problems were one layer down: in the infrastructure that lets models actually run, scale, and stay up.

So I shifted into DevOps and SRE, building Kubernetes platforms across AWS, Azure, and GCP, mostly for teams shipping AI products.

In practice: running multi-region clusters, shipping air-gapped installs for enterprise customers, and handling the security work behind ISO 27001 and SOC 2.

Lately I’m most curious about self-hosting open-weight models: serving them efficiently and running them where the data can’t leave.