About
I'm Utkarsh. My background is in ML engineering, MLOps, and cloud infrastructure. I spent years on ML pipelines and cloud infra before agents ate my whole attention span.
Now I mostly care about context engineering, memory architecture, coordination patterns, evaluation, security, deployment infra. The harness is the product.
What I Write About
The engineering behind AI agents that actually work in production. Technical deep-dives with benchmarks, comparisons, and real implementation details.
I benchmark things and write about what I find. Multi-agent patterns compared on the same task. Observability platforms tested with real cost breakdowns. No-code vs hand-coded on the same use case. That kind of thing.
Open Source
I maintain open-source repos for production agent patterns, an eval framework, and skills for Claude Code. All on GitHub.