Skip to main content

Defense Katas

Focused, repeatable verbal-and-written exercises that build fluency for the capstone defense. Do each kata in full at least once. Repeat Kata 1 and Kata 2 until they no longer surprise you.

Kata 1: 10-Minute Capstone Defense to an Imaginary Panel​

Time limit: 10 minutes delivery, 20 minutes prep the first time, 0 minutes prep by the third repetition. Goal: Deliver the capstone walkthrough out loud, with a diagram, without slides, and survive three follow-up questions. Setup: Whiteboard or paper, timer, recording device.

Deliver this structure:

  • 0-1 min -- one-line summary + outcome paragraph verbally (no reading).
  • 1-4 min -- problem and constraint setup.
  • 4-8 min -- architecture walkthrough, drawing as you speak.
  • 8-10 min -- one named tradeoff with its consequence.

After delivery, answer three prepared curveballs aloud:

  1. "What breaks first at 100x load?"
  2. "If you had to swap out your primary data store, what changes?"
  3. "Walk me through the one bug you spent longest on."

Repeat until: you can get through the 10 minutes with no filler phrases ("um," "basically," "kind of") and answer each curveball in under 90 seconds without reaching for notes.

Kata 2: Feynman-Explain 5 Topics to a Sharp Peer​

Time limit: 3-5 minutes per topic. Goal: Explain each of your five chosen Feynman topics (Cluster 5 concept 13) aloud to a peer who has not studied them. Setup: A peer (or recording). Five topics spanning Semesters 2, 4/5, 6, 7/8, 9.

For each topic, deliver the four-part structure from memory:

  1. Plain-language opening sentence.
  2. Concrete example.
  3. Visible tradeoff.
  4. Failure mode the engineer would actually notice.

After each, the peer asks one follow-up. You answer without references.

Repeat until: each of the five can be delivered with zero notes, and the follow-up question does not produce a stumble.

Kata 3: Curveball Round -- 3 Hostile Questions From a Senior Engineer​

Time limit: 5 minutes per question, 15 minutes total. Goal: Take three hostile-but-fair questions and answer each in a structured way that defends the decision without brittleness. Setup: A peer or mentor primed to be skeptical.

Example questions (generate your own three from your specialization track):

  1. "Why didn't you use ? Was that ignorance or a real choice?"
  2. "Your case study claims X works. How would I falsify that?"
  3. "Your 12-month plan seems aggressive. What's your first cut if two weeks of unexpected work hit in quarter 1?"

For each answer, aim for this structure:

  • restate the question in your own words (confirms you heard it)
  • state the decision and its main reason
  • name the alternative you considered and why you rejected it
  • state what new information would change your answer

Repeat until: you respond to each question without defensiveness and the "what new information would change your answer" sentence arrives naturally.

Kata 4: "What I'd Do Differently" Retrospective​

Time limit: 20 minutes writing, 3-5 minutes reading aloud. Goal: Write a ≤ 500-word retrospective on the capstone that names two specific decisions you would redo and one you would keep despite criticism. Setup: The capstone, your ADRs, your incident notes (if any).

Structure:

  • Two redos. For each: what you did, what you should have done, why you know that now.
  • One keeper. A choice that attracted criticism but you still defend, with the reason.
  • One thing you did not anticipate. A surprise from the project, named honestly.

After writing, read aloud. Anything that sounds like an apology, rewrite as a neutral retrospective statement.

Repeat until: the retrospective sounds like something a senior engineer would say, not an intern apologizing.

Completion Standard​

  • Kata 1 delivered twice, at least one recording retained
  • Kata 2 delivered across all five topics to at least one live listener
  • Kata 3 completed with a peer who was specifically asked to be hostile
  • Kata 4 retrospective written, edited, and filed as library/raw/retrospective.md
  • No kata feels performative; each one found at least one gap that you then closed