Episode 1. PG architecture. Day 1 of your 12 day sprint.

Scenario: the interviewer asks, "Walk me through what happens when a query hits Postgres." This is a sorting question. Answer with the mechanism path, top to bottom.

Say: "A backend process parses, rewrites, plans, then executes. Reads flow through shared buffers first, and durability comes from WAL, not the heap write."

Say the process list once: postmaster, backends, background writer, checkpointer, WAL writer, autovacuum launcher. Say it again until it is clean.

Mechanism: shared buffers cache pages, WAL writer appends sequentially, checkpointer bounds recovery time. Say: "Checkpoints trade recovery time against IO spikes."

Lab: install Postgres locally, run a pgbench smoke test, then query pg_stat_activity and name every backend state you see.

Bridge from Oracle: Say: "Where Oracle has SGA plus redo, Postgres has shared buffers plus WAL. Same durability contract, different writer layout."

Drill: pause and recite the six processes from memory.
Say: postmaster, backends, background writer, checkpointer, WAL writer, autovacuum launcher.
Say the durability line once more: durability comes from WAL, not the heap write.
Done-check: name all six plus the checkpoint trade-off before bed.
Depth tag tonight: Debug. You ran pgbench and read the activity view yourself.

Teaser: tomorrow, Episode 2, MVCC. Why readers never block writers, and the bloat bill that comes due.