Exercises: The Data Engineering Career

Most of these are about you, and several will be uncomfortable. Exercises marked 🧪 use code/career_map.py.

Do Exercise 40.1 today, not when you finish the chapter. It is the only one with a deadline set by your own memory.

Start Now

Exercise 40.1 — Start the running document

Difficulty: ★☆☆ · Time: 5 minutes, then 90 seconds a week

§40.3 and §40.13's 📐.

  1. Create the file. Anywhere. Two columns: date, and what happened.
  2. Write three entries from the last month — a design you improved, a question that changed something, a thing you talked someone out of. You will find this harder than you expect, which is the point.
  3. Set a rule: anything that makes you think "that was useful" goes in within the hour, without judging whether it counts.
  4. In two weeks, count the entries. How many would you have remembered without it?

Exercise 40.2 — The ten-second test

Difficulty: ★☆☆ · Time: 20 minutes

§40.11: "if this product disappeared tomorrow, what would I still know?"

  1. List your five deepest technical skills.
  2. Apply the test to each. Write down what survives — specifically, not "the fundamentals."
  3. Which of the five has the highest product content? Case Study 1's engineer was 30% product, 70% properties. What is your ratio?
  4. §40.11 says the gap — knowing things you cannot articulate outside their product context — is the whole vulnerability. Find one of yours and name it.

Exercise 40.3 — Which question are you asking?

Difficulty: ★☆☆ · Time: 15 minutes

Case Study 1's finding: two people, the same four years, different outcomes, decided by which question they were asking.

  1. Recall your last three debugging sessions. Were you asking "how does this tool behave?" or "why is this slow / wrong?"
  2. Be honest about the ratio.
  3. §"Read the Plan" says you will not find out which you chose for several years. What earlier signal could you use?

Levels and Scope

Exercise 40.4 — Place yourself

Difficulty: ★★☆ · Time: 30 minutes · 🧪

python code/career_map.py --levels
  1. Which question are you currently trusted to answer? Not which you can answer — which you are trusted to.
  2. Read the "judged on" column. What were you actually judged on in your last review?
  3. §40.3's four kinds of senior evidence: how many do you have from the last year?
  4. A thing you did not build, with a written reason — do you have one? It is called the strongest artifact and almost nobody has one.

Exercise 40.5 — Ask for an instance

Difficulty: ★★☆ · Time: one meeting

Case Study 2's four-minute question.

  1. Ask your manager: "Can you show me something a person did that got them promoted here? Not the criteria — an actual thing somebody did."
  2. Get three examples if you can.
  3. What shape do they share? Case Study 2's were all about something not happening.
  4. Compare against your written criteria. How different are the two documents?

Exercise 40.6 — Does the senior work exist?

Difficulty: ★★☆ · Time: one request

Case Study 2's 🏭 test.

  1. Take the next request that arrives and ask what decision it serves.
  2. Was the question welcomed, tolerated, or treated as obstruction?
  3. Score your team against §"From the Pipeline"'s two columns.
  4. If you are in the second column: §40.3 says that is a reason to leave rather than to try harder. Do you agree? What would change your mind?

Specialization and Teams

Exercise 40.7 — Score the tracks for yourself

Difficulty: ★★☆ · Time: 40 minutes · 🧪

python code/career_map.py --tracks
  1. Rescore all five tracks with your own numbers, for your market and your preferences.
  2. Check that no track dominates after your changes. If one does, you have expressed a preference rather than a trade-off — which is fine, but say so.
  3. §40.5 says pick the column you are least willing to compromise on. Which is it?
  4. Is your current track the one that column implies?

Exercise 40.8 — Score your own team

Difficulty: ★★☆ · Time: 30 minutes · 🧪

python code/career_map.py --health
  1. Score your team honestly. Twenty signals, range −31 to +39.
  2. Which of the +4 and +5 signals do you have? Those are the ones that cannot be bought.
  3. §40.9 says good signals require sustained effort and bad ones are the resting state. Does your score support that?
  4. Pick the highest-weight signal you are missing and cost out getting it. For most teams that is the reconciliation.

Exercise 40.9 — Score a team you are interviewing with

Difficulty: ★★☆ · Time: one interview

  1. Take the scoring sheet into a real conversation, using Chapter 39 §39.10's questions.
  2. Score during, not after.
  3. Set your decline threshold before the call. Chapter 39's Exercise 39.14 puts it plainly: a list with no decline rule is a list whose answers you rationalize.
  4. Compare your score against your gut feeling. Where did they disagree, and which was right?

Getting Better

Exercise 40.10 — Audit your learning

Difficulty: ★★☆ · Time: 30 minutes

§40.10's compounding and non-compounding activities.

  1. List what you did to get better last year. Courses, reading, side projects, conference talks, on-call, postmortems, reviews.
  2. Classify each: attached to a system you were responsible for, or not.
  3. What fraction was attached? §40.10 argues only the attached kind compounds.
  4. §40.10's one-sentence version: ask to own something you are slightly not ready for, and make sure it pages you. What would that be, concretely, this quarter?

Exercise 40.11 — Take the unwanted job

Difficulty: ★★★ · Time: a quarter

Case Study 1's 🏭: "if your expertise is entirely upstream of meaning, take something downstream of it."

  1. Can you name a number your business makes a decision with? If not, that is the finding.
  2. Volunteer for the thing nobody wants — the reconciliation, the metric definitions, the finance close, the data quality register.
  3. Do it for a quarter. Write down what you learn about the business that you did not know.
  4. Case Study 1's engineer says this was the actual lesson of a decade. Was it, for you?

Advanced

Exercise 40.12 — Extend the map

Difficulty: ★★★ · Time: 90 minutes · 🧪

Add three to career_map.py, with self-checks:

  1. A sixth track for your market, scored — and check that it does not dominate.
  2. A self-assessment mode that takes your answers to the twenty health signals and scores them.
  3. A level-evidence checklist from §40.3's four kinds, with a place to record instances.
  4. A skill-depreciation calculator: for each skill, a product fraction and a years-since estimate.
  5. A "questions I asked" tracker for Chapter 39 §39.10, so a job search accumulates comparable scores.
  6. A weighting sensitivity check: which single health signal most changes a team's verdict?

Then run #6. If one signal dominates, the twenty-signal list is that signal with decoration — say whether that is a problem.

Exercise 40.13 — Write your own version of §40.1

Difficulty: ★★★ · Time: 45 minutes

§40.1 claims the job has changed shape three times and the core has not moved.

  1. Write the equivalent for your own career, in the same format: what the job was, in three or four eras.
  2. What has been constant? Be specific; "problem solving" does not count.
  3. If you cannot find a constant, that is worth knowing — it may mean you have been changing jobs rather than deepening one.
  4. §40.1 says the constant is "somebody has to say what a number means and whether it is right." Do you agree that is the durable core of this field? Argue against it.

Exercise 40.14 — The five-year letter

Difficulty: ★★★ · Time: 60 minutes

Write a letter to yourself, to be read in five years.

  1. What you currently believe is durable in your skill set, and why.
  2. What you suspect is perishable, and what you are doing about it.
  3. The track you chose and the column you optimized for (§40.5).
  4. The team-health score of where you work now (§40.9), and what you decided about it.
  5. One prediction about the field that you are willing to be wrong about in writing.

Put a calendar reminder on it. Case Study 1's engineer spent a year believing four years were wasted; a letter from 2013 would have shortened that considerably, because it would have shown them what they were actually learning.

The Last One

Exercise 40.15 — How do you know?

Difficulty: ★★★ · Time: the rest of your career

§40.13.

  1. Pick the number your organization cares about most.
  2. Trace it to its source. Every rule, every transformation, every assumption.
  3. Ask, at each step: how do I know this is right? Not "does it look right."
  4. Write down every place the honest answer is "I don't."
  5. Fix one of them.

That list is the job. Everything in these forty chapters — the partitioning, the idempotency, the assertions, the contracts, the reconciliation, the layer rules, the deletion manifest, the cost model — exists to shorten it.

And the reason it matters is the only thing worth carrying out of this book: you are the reason a number on a screen means what someone thinks it means, and nearly nobody downstream is in a position to check.