Daily Log

Sent 29 applications — kept the volume-game pace maxed out.

• ✓ Sent 29 applications

Traced the auth request lifecycle across three machines — the session check runs on the Vercel server, and FastAPI is only reached after it verifies.

A method for learning an unfamiliar codebase fast: generate a why-first, self-checked reading artifact instead of reading files cold.

Daily study + job-search day — detailed notes kept private.

• ✓ Sent 15 applications

Daily study + job-search day — detailed notes kept private.

📝 Wrote notes: resume deep-dive (Ownership / Tech Stack / Impact) + two-layer resume strategy.

Daily study + job-search day — detailed notes kept private.

Sunday report-style log — kept the application cadence going.

• ✓ Reviewed application materials
• ✓ Sent 18 applications

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

• ✓ Sent 15 applications

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

• ✓ AI project Day 2: MVP implemented
• ✓ Backend project Day 3: PostgreSQL
• ✓ Sent 15 applications

Backend MVP: design a couple of endpoints, ship the sync version first, then understand it by tracing one request — and read code in zoom levels, not line by line.

• ✓ Backend project Day 2: sync API + Docker MVP
• ✓ Sent 10 applications

📝 Reading code at three zoom levels (big picture first); config management via .env → environment variables → code; how to write a project résumé bullet

Sliding-window min-length (LC 209) is O(n) because the left pointer never backtracks; and an accumulator's init value must be the operation's identity (sum→0, product→1).

• ✓ 0x3f Ep.01 framework summary
• ✓ 0x3f Ep.02 framework summary
• ✓ System Design #2: API Design

📝 LC 209 sliding window (caterpillar, same-direction two pointers, amortized O(n)); accumulator init = the operation's identity (sum→0 / product→1 / max→−inf), a lesson from LC 713

Read a problem's constraints as a budget (~10⁸ ops/second) and plug N into each complexity to pick the right approach before coding.

📝 Constraints → target complexity (≈10⁸ ops/second; infer the approach from N's magnitude); computing time/space (count how many times each element is processed + audit allocated variables)

Daily study + job-search day — detailed notes kept private.

• ✓ Backend project Day 1 done: MVP implemented

A lighter day — kept the core study cadence; detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

• ✓ AI project Day 1
• ✓ Backend project Day 1
• ✓ 0x3f blind-write confirmed 16 / 18 / 611 — Ep.01 all 6 cleared

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

📝 0x3f Ep.01 homework: rewritten after reading the solution; blind-write still to confirm

A lighter day — kept the core study cadence; detailed notes kept private.

📝 Interview: three sentences to align on the approach with the interviewer and confirm edge cases before coding

Daily study + job-search day — detailed notes kept private.

System Design basics: "stateless" describes the server (nothing in its RAM the next request needs) — the database holds the truth, servers stay disposable.

• ✓ System Design #1: Client/Server + Stateless

📝 Externalize state to Redis/DB to keep the server stateless; the 3 System-Design interview triggers + a 4-step answer (concede → classify → cost → mitigate)

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Opposite-direction two pointers: on a sorted array, finding a pair is O(n) instead of O(n²) — move the pointer that can improve the result.

📝 Opposite-direction two-pointer template (move the end that can improve the result); LC 42 Trapping Rain Water — prefix+suffix idea + two-pointer O(1)-space optimization

Daily study + job-search day — detailed notes kept private.

The HashMap pattern (Two Sum, Group Anagrams, Longest Consecutive) trades space for time to turn O(n²) lookups into O(n).

• ✓ Rewrote 6 problems from memory in 21 min

📝 English rewrite of the HashMap pattern (Two Sum, Group Anagrams, Longest Consecutive) + Prefix Product; when to skip else; oral-English phrases

A hash map is O(1) because the hash function computes the index directly (no search), degrading to O(n) only when collisions chain a bucket.

📝 Hash Table: any traversal is O(n); why Hash Map is O(1) (the hash function computes the index directly) and how collisions degrade it to O(n)

Daily study + job-search day — detailed notes kept private.

Daily study + job-search day — detailed notes kept private.

Counting with a dict plus the prefix/suffix trick (LC 242 / 347 / 238) brings O(n²) down to O(n).

📝 HashMap counting (242 Anagram) + Top K (347 sort-by-value template) + prefix/suffix products (238, O(n²)→O(n))

Daily study + job-search day — detailed notes kept private.

List vs Set vs Dict — choose by the question you need to answer, and mind the Python gotcha that {} is an empty dict, not a set.

📝 List vs Set vs Dict — core differences and complexity; the O(n log n) sorting lower bound; Hash & Complexity interview lines + a general Hash Set template

Hashmap basics via Two Sum — a dict's O(1) lookup replaces a list's O(n) (space for time) — plus dict key rules and an edge-case checklist.

📝 Hash map basics: dict/set O(1) lookup vs list O(n), trading space for time; Two Sum takeaways + edge-case checklist (empty / single element / duplicates / negatives / no solution)