Week 01 — Baseline + foundations

Theme: Find out where you actually are, then rebuild the primitives. Budget: ~14 hrs. New problems: 18 core + 3 stretch (marked ☆). Exit criterion: you can write a correct binary search that handles duplicates — first pass, no debugging.

Every problem: plain doc, timer running, spoken out loud, no autocomplete, no running the code. Log every attempt in 05-practice-log.md the moment you finish. Non-negotiable.

Day 0 — Setup (30 min, before Monday)

• Create a Google Doc named interview-scratch. Set font to Consolas or Courier New, 11pt. Turn off spellcheck and autocorrect (Tools → Preferences → uncheck automatic substitution). This is your only coding surface for 8 weeks.
• Put a visible countdown timer on a second screen or phone. Not a stopwatch — a countdown, so time pressure is felt.
• Decide your language and commit to it. Whatever you’re fastest in. Do not switch mid-prep.
• Write your language’s cheat-sheet page in the doc, from memory: how to declare a hashmap/set/heap/deque, sort with a custom comparator, and do integer division. You will fumble these under pressure otherwise.
• Open 05-practice-log.md and keep it open.

Day 1 (Mon) — Cold baseline · 2 hrs

Do not look anything up. Do not skip a problem because it looks hard. The point is an honest measurement, not a good score. A bad baseline is useful data; a dishonest one wastes eight weeks.

Timed block — 75 min (25 min each, hard stop):

Review block — 45 min:

• For each: did you solve it unaided in 25 min? Was the code correct before debugging?
• For each miss, write the specific reason in the log. Not “I forgot DP” — rather “I couldn’t state the recurrence; I jumped to a table.”
• Only now look up the intended solutions. Re-read yours side by side.
• Add every miss to the Redo queue. They come back on Day 7.

Calibration: 3/3 unaided → you’re closer to Week 3, tell me and I’ll re-scope the plan. 1–2/3 → this plan is correctly sized. 0/3 → still fine, but expect Weeks 1–4 to run hot; consider 12 weeks.

Day 2 (Tue) — Binary search from first principles · 2 hrs

Binary search is the highest ratio of asked to reliably-written of any primitive. Most engineers write it 90% correct, which is 0% correct.

Derivation block — 45 min. No reference material.

• Write three separate functions and name them properly:
  • find_exact(a, t) → any index of t, or −1
  • lower_bound(a, t) → first index with a[i] >= t
  • upper_bound(a, t) → first index with a[i] > t
• Use mid = lo + (hi - lo) // 2 and be able to say why.
• Write down your loop invariant in words above each function. If you can’t state the invariant, you don’t have it memorized — you have it half-memorized, which is the dangerous state.
• Hand-trace all of these on paper:

[]              t=1     empty
[5]             t=5     single, hit
[5]             t=3     single, miss low
[5]             t=9     single, miss high
[1,2]           t=2     two elements, upper
[2,2,2,2]       t=2     all duplicates, present
[2,2,2,2]       t=3     all duplicates, absent
[1,3,5,7]       t=0     below range
[1,3,5,7]       t=9     above range

Problem block — 75 min:

☆ Stretch: Find Minimum in Rotated Sorted Array II (154) — the duplicates case that breaks the naive invariant.

Day 3 (Wed) — Two pointers + prefix sums · 2 hrs

Problem block — 100 min:

Recall block — 20 min:

• Close everything. Rewrite #9 (Subarray Sum Equals K) from memory, start to finish. Rewriting from memory 24 hrs later is what moves a pattern into long-term storage — reading the solution again does almost nothing.

Day 4 (Thu) — Sliding window · 2 hrs

Derivation block — 20 min:

• Write the generic variable-window template in your own words — expand right, contract left while the invariant is violated, record the answer. One template covers every problem below. Write it once, then instantiate it four times.

Problem block — 100 min:

☆ Stretch: Permutation in String (567) — fixed window, cheap win.

Day 5 (Fri) — Behavioral brain dump · 1 hr

Raw material only. No STAR formatting, no polishing. You cannot write good stories from a blank page in Week 6; you write them from a list you made in Week 1.

• In 04-behavioral-story-bank.md, list every project, incident, migration, launch, and disagreement from the last 3 years. Target 12–18 raw entries.
• For each, one line each, fast:
  • What was it?
  • What did I specifically own?
  • What was the hardest decision (not the hardest task)?
  • Who disagreed with me, and what happened?
  • Any number attached to the outcome?
• Star the 3 with the largest blast radius beyond your own team — those become your anchor stories.

If you’re drawing a blank: scan your calendar, PR history, and Slack/email from the last 3 years. Memory is a terrible index.

Day 6 (Sat) — Mixed timed set + primitives · 3 hrs

Timed set — 75 min (25 min each, mixed categories, no hints). Mixing is the point: real interviews don’t tell you the topic.

Primitives block — 60 min. From memory, no reference:

• Lomuto partition and Hoare partition. Know which one quickselect prefers and why.
• Quickselect → solve Kth Largest Element in an Array (215) with it. State the average O(n) and worst O(n²), and how random pivoting fixes it.
• Iterative inorder traversal with an explicit stack — Binary Tree Inorder Traversal (94).
• Iterative preorder — Binary Tree Preorder Traversal (144).

Weekly review — 45 min:

• Fill in the Weekly review block in 05-practice-log.md.
• Compute your two numbers: unaided solve rate and first-pass correctness.
• Name your single weakest area in one sentence, and one concrete change for Week 2.

Day 7 (Sun) — Spaced repetition · 2 hrs

Cold re-solves. This is the highest-value block of the week, and the one people skip. Do not replace it with new problems.

• Re-solve the three Day 1 baseline problems (15, 200, 322) cold, timed at 25 min. Compare against Monday. The delta is your real progress signal.
• Re-write all three binary search variants from memory again. Re-run the 9 edge cases from Day 2 by hand.
• Clear the redo queue: anything else you failed this week, re-solve now.
• Preview Week 2: skim the estimation and non-functional requirements sections of 03-system-design-curriculum.md. 10 min, no notes. Priming only.

Week 1 exit checklist

• Baseline recorded honestly for all 3 problems
• All 3 binary search variants written from memory, correct, first pass, including duplicates
• 18 core problems attempted and logged
• Every failed problem re-solved at least once on Day 7
• 12+ raw behavioral entries captured
• Practice log filled in for every single session — no blank rows

If you missed the binary search criterion: do not move on to Week 2’s new topics on Monday. Spend the first 45 minutes of Week 2 Day 1 re-deriving it. It compounds into Weeks 3–4 and it is genuinely cheap to fix now.