The governing rule
Choose a project with three consequential decisions, one failed approach, defensible evidence, and an outcome you can bound. A famous project you barely owned is weaker than a smaller project whose hard calls were yours.
Build the deck around three decisions
- Open with the thesis. State the problem, stakes, your ownership, and measured outcome in four sentences.
- Give only necessary context. The audience needs users, constraints, prior behavior, and failure cost, not a company history.
- Defend each decision. Show alternatives, evidence available at the time, the choice, and what would have changed it.
- Descend once. Move from architecture into one model, bottleneck, failure trace, or derivation that proves hands-on depth.
- Show the wrong turn. Name the assumption, disconfirming signal, and response.
- Bound the impact. Separate observed outcome, estimated contribution, team output, and second-order effects.
12-slide outline
A decision-centered deck structure and slides to delete.
Challenge bank
Twenty skeptical follow-ups across evidence, implementation, ownership, and counterfactuals.
Run the simulation
Thesis and ownership
02–07Problem, constraints, prior state
07–23Three decisions and one technical descent
23–27Failure, evaluation, operations
27–30Impact, changed judgment, summary
30–45Interrupted Q&A
Score before coaching
- Thesis: the opening establishes stakes, claim, ownership, and result.
- Decision spine: alternatives and evidence organize the narrative.
- Technical depth: one implementation detail survives expert follow-up.
- Evidence: metrics, slices, uncertainty, and attribution are honest.
- Defense: direct answers come before context; the candidate updates when warranted.
- Level: scope and influence match the target without replacing hands-on credibility.
Practice gate
Run the presentation twice. The second observer must not have seen the deck, must interrupt, and must ask at least one counterfactual. Do not graduate if the story loses ownership, evidence, or technical credibility under questions.