Low-Level Design (LLD / OOD)

How to approach a 45-minute LLD round: the 6-step framework, scoping questions, when to code vs diagram, anti-patterns, and how interviewers evaluate.

The GoF patterns re-grouped by the LLD trigger that should make you reach for each one — Strategy, Factory, Builder, Observer, State, and the rest.

When the LLD round expects thread-safe code, and the toolkit: synchronized, ReentrantLock, ReadWriteLock, ConcurrentHashMap, atomics, BlockingQueue, and immutability.

The most common LLD problem worldwide. Multi-floor lot, typed spots and vehicles, ticketing, pricing strategy, and availability notifications.

Scheduling strategies (FCFS, SCAN, LOOK, nearest-car), elevator state machine, button commands, and multi-car controllers.

Books vs book-items (copies), members, loans and reservations, fine calculation, and overdue notifications.

The classic State-pattern problem: Idle / HasMoney / Dispensing / OutOfStock states, payment strategy, and inventory.

Polymorphic piece movement, board modeling, move validation per piece, undo via Command, and the game state machine.

Board, snakes/ladders, dice strategy, players, and the turn loop — a compact warm-up problem.

ATM state machine, transactions as commands for logging/replay, denomination dispensing via Chain of Responsibility, and bank integration.

Cinemas, shows, seats, and the central concurrency challenge: two users booking the same seat. Optimistic vs pessimistic locking and seat-hold TTL.

Rider/driver model, ride state machine, surge-pricing and matching strategies — the LLD slice of a hybrid LLD/HLD problem.

The pattern-composition showcase: Chain of Responsibility, Strategy, Singleton, Observer, and Decorator combined into one logging framework.

Token Bucket, Leaky Bucket, Fixed Window, Sliding Window Log/Counter — the algorithms in depth, behind a Strategy interface. The highest-leverage senior problem.

O(1) LRU via HashMap + doubly-linked list, and the harder O(1) LFU via frequency lists plus a minFreq pointer. Eviction as a Strategy.

Broker, topics, publishers/subscribers, concurrent delivery, delivery semantics, and backpressure — the in-memory precursor to Kafka.

Encoding strategies (base62 vs hash vs counter), the class model, and collision handling. Scaling is deferred to the HLD module.

Channel abstraction (Email/SMS/Push/InApp), templates, retry policy, and channel fallback via Chain of Responsibility.

The Composite-pattern problem: directories contain files and directories. mkdir/touch/ls/cd/rm/mv/cp over an in-memory tree.

Clocks, timers, scheduled tasks, and scheduling strategies (FIFO, priority, cron) with tasks modeled as commands.

How many problems to solve, the internalizable template, 45-minute time-boxing, common mistakes, and how to signal senior.