Unofficial. A job application for OpenAI’s Applied AI Delivery Lead role, built from public pages. Not affiliated with OpenAI.

The delivery health board.

The plan’s first deliverable, as a page you can use. You are in a delivery review for one deployment. Pick a review date. Click any gate, risk, workstream or number to see why it has the status it has.

Start here

Why I did thisYour posting asks for an integrated delivery plan and a clear view of delivery health. Those are easy words to agree with. I wanted you to see the actual review I would run.

What I was thinkingOpenAI publishes most of what a customer needs to plan a rollout, spread over several sites. Nobody outside can see it joined up for one deployment. Joining it up, with an owner for every piece, is the job.

What it showsA status board for one rollout at a made-up insurer. Every gate, risk and number can be clicked to show who owns it, why it has that status and where the figure came from.

Try it in four steps

You are in the weekly delivery review.

  1. Pick “Week 10, go or no-go” at the top. This is the week the launch decision is due.
  2. Look at “Launch confidence”. It shows a dash, not a colour, because one gate has no data behind it.
  3. Click gate 8. The panel on the right says who owns it, why it is blank and what would change it. Then record what the owner reports.
  4. Click any risk or number. Each one says where it came from, and whether a person or a system supplied it.

Why I am the person for this job. This is my day job at a different scale: I run delivery for ten vendor teams with contracted scope and a launch date that does not move, and I built the board they plan from. I have led delivery from the client side, not from inside a model provider. The habits carry over: one owner each, every number sourced, nothing green that nobody checked.

delivery health · prototype Sample data. Fernhill Mutual is made up. Not an OpenAI product.
Claims assistant · Fernhill MutualA made-up insurer. Launch planned for week 11 of 16.

Gates

Workstreams

Risks

Each risk has one owner and the gate it threatens.

Adoption against the baseline

MeasureBaseline, week 1NowWhere the number comes from

What a moved date means for usage

Bars are sample usage units per week, with no scale on purpose. The shape is the point: the ramp starts when the launch does.

What this replaces

Most delivery reviews run from a slide deck that one person updates by hand the night before. A green light on a slide says nothing about who checked it or when.

I have no view of how OpenAI runs its reviews today, so this replaces nothing of theirs. It shows the review I would bring.

Choices I made in this page

  • A gate with no data shows a dash, never green.At week 10, one gate has nothing behind it, so the board states no launch confidence until a person supplies the figure. This rule comes from my own planning board.
  • Every number carries its source.Stated in a system, joined, derived, or entered by a person. Self-reported hours are marked as estimates, as OpenAI’s own guide asks.
  • Launch confidence is derived, and says from what.It is a rule over the gates, not somebody’s mood on the day.
  • Facts from OpenAI’s public pages are marked and linked.Everything else on the board is sample data.
  • The usage picture has no dollar figures.I have never owned a revenue number, and I will not draw one I cannot source.