Nicholas Dunzelman

AI coding deployment field kit ยท checked May 19, 2026

Templates for turning an AI coding demo into deployment evidence.

This is the operational layer behind the proof packet: intake, scope, receipt, failure taxonomy, decision table, and product feedback. It is built so a second engineer can inspect the run after the call ends.

Kit

Six artifacts

Intake Five questions decide whether a task is safe enough and useful enough to demo.
Scope card Owner, task, verifier, data boundary, rollback, and human review rule before the agent runs.
Receipt Changed files, commands, visual proof, verifier result, failure, decision, and replay path.
Taxonomy Six buckets separate acceptance, context, harness, environment, tool, and review-boundary misses.
Decision table Ready, repair, or reject for each task pattern before it becomes a team habit.
Feedback packet Field misses become product signal instead of transcript residue.

Runbook

How the kit is used

  1. 01

    Intake before demo

    Reject tasks with no owner, no falsifiable check, broad blast radius, hidden credentials, or ambiguous review rules.

  2. 02

    Scope card before prompt

    Name the repo, owner, acceptance check, verifier, rollback, data boundary, disallowed actions, and reviewer.

  3. 03

    Receipt after run

    Capture context loaded, files touched, command output, browser proof, failure observed, and replay command.

  4. 04

    Decision before rollout

    Classify each pattern as ready, repair, or reject. Do not scale from excitement alone.

  5. 05

    Feedback into product

    Translate misses into model behavior, tool harness, docs gap, repo context, permission boundary, or customer-readiness signal.

Fit

Why this matters

The AI Deployment Engineer - Codex posting asks for customer workflow design, demos, reference implementations, workshops, technical content, product insights, SDLC strategy, security considerations, and operational readiness.

The field kit turns those nouns into artifacts a hiring manager can inspect. It does not claim customer deployment history; it shows the operating shape Nic would bring into a deployment lane.

Proof stack

Filled sample run Receipt reference Proof brief 30-day plan Workshop kit Patterns guide Cleanroom repo

Boundary

Cleanroom line

This page is independent from OpenAI. It uses OpenAI and Codex only to name public role context. It does not use OpenAI logos, product UI, private systems, or affiliation language.

Sources checked: AI Deployment Engineer - Codex and OpenAI interview guide.