Sandeep Mehta
Agentic engineering, from the work.
I have spent more than a decade building distributed systems, data platforms and streaming systems. My work now focuses on applied AI: how agents are designed, evaluated, controlled and operated when people depend on the result.
11factor is my framework and public working notebook for agentic engineering: engineering with AI agents and building reliable agent-based systems. I write about clear tasks, context, tools, orchestration, evaluations, observable execution and permissions. I want evidence for each improvement, and keep the questions open when the work does not yet support an answer.
The work
The framework connects tasks, controlled execution and measured improvement. Projects are bounded examples. My setup describes the tools I use. The writing is where I share what I learn and believe.
Notes and experiments on agentic engineering: tasks, context, tools, evaluations and what the evidence supports.
→Oyako and Clinly, with explicit limits on what their current evidence proves.
→My framework for engineering with AI agents and building reliable agent-based systems: Build, Run, Observe and Improve, with controls throughout.
→Hermes Agent, Claude Code and Codex — for task work and deeper testing.
→Publishing boundary
I write from public sources and work in my own lab. Employer-confidential material, customer information and private operating details stay out of these pages.
A system page explains the current method and its limits. It does not turn a local result into a claim about security, scale or customer value.
Follow the writing and working notes.