What Makes Someone Good at This
7+ years of engineering experience, applied at a higher altitude. You need years of building and debugging production systems. Not because you’ll write every line, but because you can’t design a harness that catches real failures, write a spec that anticipates edge cases, or diagnose a broken feature across the full stack without that foundation. The depth serves the abstraction.
Systems thinking over code fluency. How components interact. Where failures cascade. What breaks when requirements change. What to anticipate before it happens. This is what agents are worst at and what matters most.
An agent-driven workflow. You already direct AI agents (Claude Code, Codex, Cursor, or similar) to handle implementation while you focus on architecture, specification, and validation. Or you have the engineering judgment to make that transition and the motivation to do it now.
Experience building the infrastructure around agents. CI enforcement, scenario-based testing, documentation systems agents can consume, structured knowledge bases — you’ve built some of this, or you have specific ideas about how and why.
Comfort making decisions with incomplete information. Startup. Requirements shift. The right approach isn’t always obvious. You move forward, and you know when to ask versus when to make a call.
Direct communication. You give and receive honest feedback. You can disagree with a decision, say so clearly, and still commit to the outcome. We care about getting it right more than being right.
Enthusiasm for a field that reinvents itself quarterly. Tools change. Workflows get replaced. Best practices from three months ago become obsolete. You’re energized by that. You see this as the most interesting period in the history of software.