← Work index
CASE 01 / Technical operationsUSED

Dynamo Air

Keeping engineering information reviewable

AI-assisted documentation inside an aerospace engineering environment. Source control and review made changes inspectable; engineers remained the technical authority.

CONTEXT
Professional contract
TOOLS / METHODS
Git · Markdown · Claude Code · Review workflows
THE CENTRAL LESSON

AI can accelerate the production of technical material. The harder work is keeping assumptions explicit and making every meaningful change reviewable by someone with the authority to judge it.

1 evidence recordRead the scope & limits ↓
THE QUESTION

How do you preserve technical correctness when information is changing, distributed, and AI-assisted?

Context

Technical documentation becomes part of an engineering information system when other people rely on it to make decisions. My work at Dynamo Air involved drafting and revising material in a Git-based Markdown repository, using AI assistance while keeping engineering review central.

What I did

  1. 01

    Worked with repository-based technical documents rather than isolated generated drafts.

  2. 02

    Used Claude Code, Markdown, branches, commits, and reviewable changes to develop and revise material.

  3. 03

    Reviewed consistency across assumptions and references, then incorporated engineering feedback.

Architecture / workflow

DYNAMO AIR / SYSTEM STUDYREPOSITORY + ROLE ACCOUNT

A draft is a proposal. Review gives it context.

Who judges technical correctness?

  1. 01
    INFORMATION / ARTIFACT

    Technical sources

    Documents and engineering feedback

  2. 02
    MODEL ASSISTANCE

    Assist the draft

    AI-assisted writing and consistency review

  3. 03
    SOFTWARE LOGIC

    Expose the change

    Markdown, Git branches and reviewable diffs

  4. 04
    HUMAN JUDGMENT

    Engineer review

    People retain technical authority

INSPECT THE FAILURE BOUNDARYA plausible draft contains a stale reference or incompatible assumption.
WHAT FOLLOWS

Review must examine the changed claim and its source. Version history alone cannot establish correctness.

An abstraction of the documentation workflow. No aircraft design, certification authority, or proprietary document is depicted.

Where it gets difficult

A fluent draft can conceal incompatible assumptions. The useful unit of review is the changed claim and its source, not how convincing a paragraph sounds.

Source control makes differences visible, but it does not establish technical truth. Review needs both a traceable change and the right technical reviewer.

THE BOUNDARY OF THE CLAIM

Limits & evidence

  • I am describing documentation and review work, not aircraft design authority or certification responsibility.
  • Company engineering documents and technical specifications are withheld. The public account describes the process, not the aircraft.
  • Repository activity confirms participation; it does not establish every employment date or the downstream effect of a particular correction.
Source inspection

Source-controlled engineering documentation

Local repository and authored commit history support participation in Git-based technical documentation. The supplied account and prior portfolio review describe AI assistance and engineering review.

What this does not establish

Company documents remain private. Repository participation does not establish design authority, certification responsibility, exact employment dates, or measured outcomes.

What this changed

AI can accelerate the production of technical material. The harder work is keeping assumptions explicit and making every meaningful change reviewable by someone with the authority to judge it.

WHERE THIS LEADS

This led toward a broader question in applied AI: what should the model propose, and what must the surrounding system verify?

NEXT CASEConduit