Low Level Design — OOP, UML & Class Relationships

Section I — Notebook Summary

═══ WHY LLD ═══
LLD = Low-Level Design → detailed blueprint BEFORE coding.
Build-a-house analogy:
  jump in       → inconsistencies, rework
  high-lvl only → vague, confusion
  detailed plan → BEST: roadmap, aligned team, fewer errors ✓
Gives: exact specs to start coding.
Goal of good design → scalable + low-maintenance + robust.
Design on paper first → know where dev is going.
Expected from devs with > 1 yr experience.

═══ TYPES OF LLD INTERVIEWS ═══

recall: Theory→talk, Design→sketch+some code, Machine→ship it.

═══ PREREQUISITES ═══
- OOP lang exp (Java/C++/Python)
- OOP concepts: Encapsulation, Inheritance, Abstraction, Polymorphism
                + Association / Aggregation / Composition   (this session)
- SOLID principles            (later sessions)
- Design patterns: bonus(junior) / expected(senior)  (later)

═══ 4 PILLARS OF OOP (term → goal) ═══
Encapsulation → reduce complexity + data security  (bundle data+methods, hide fields)
Abstraction   → hide complexity + isolate impact   (expose what, hide how)
Inheritance   → kill redundant code + reuse        (is-a)
Polymorphism  → one object, many forms             (override / overload)

═══ HOW TO REPRESENT LLD ═══
Words (email/meeting/Slack) → ambiguous, "I didn't mean that", hard for complex.
Diagrams → 1 pic = 1000 words → less ambiguity, easier to visualize. ✓
UML = Unified Modelling Language → STANDARD way to draw software concepts.

═══ UML CLASS DIAGRAM (1 box, 3 compartments) ═══
┌────────────────────────┐
│      ClassName         │   ← name
├────────────────────────┤
│ - id : int             │   ← attributes (fields)
│ - name : String        │
├────────────────────────┤
│ + getName() : String   │   ← methods
│ + setName(String):void │
└────────────────────────┘
Visibility:  + public   - private   # protected   ~ package
Syntax:  attr →  vis name : Type        method → vis name(params) : Return
underline = static, italic = abstract

═══ CLASS RELATIONSHIPS (the "has-a" family) ═══
Association ⊃ Aggregation ⊃ Composition   (each a stricter special case)

Association  → Student ↔ Teacher        (use each other, no owner)
Aggregation  → Car ◇── Passenger        (owner, but passenger lives on)
             → Department ◇── Professor  (delete dept ⇒ profs still exist)
Composition  → House ◆── Room           (delete house ⇒ rooms gone)
             → Order ◆── OrderLineItem   (line items have no life of their own)
test: "if I delete the parent, must the child die?"  yes→Composition  no(but owned)→Aggregation  no owner→Association

═══ CARDINALITY (how many ↔ how many) ═══
Definition: # of instances of one class that relate to instances of another.

UML notation: 1, 0..1, *, 1..*, 0..*  written at each end of the line.

Section II — Detailed Study Notes

1.Why Low-Level Design?

Imagine being asked to build a house. There are three ways to start. Jump right in and build without a plan — you make decisions on the fly and risk inconsistencies and structural problems. Do only a high-level design — a rough blueprint of the overall shape — and you'll hit confusion and errors because the specifics were never nailed down. Or do detailed planning first: exact materials, measurements, and techniques, every aspect decided before the first brick. The third path wins. It gives a precise roadmap, aligns everyone on their role, and minimizes errors and rework, producing a structurally sound result.

Software is the same. LLD is the equivalent of those detailed construction plans for a module: which classes exist, what they hold, how they relate, and how they collaborate. You design on paper before jumping into code so you always know where the development is heading.

+ A useful mental model: HLD (High-Level Design) answers "what are the big boxes and how do they talk?" (services, databases, queues), while LLD answers "inside one box, what are the classes, fields, methods, and relationships?" They sit at different zoom levels — LLD is the close-up.

2.What LLD Actually Delivers

The point of building scalable, low-maintenance, robust software is something every engineer is expected to care about, and LLD is how you get there deliberately instead of by luck. It is good practice to design modules on paper before coding because it surfaces gaps early and keeps the development process pointed in a clear direction. Once you have more than about a year of experience, some level of LLD fluency is simply expected of you.

+ Quick definitions of the three target qualities, since they're the whole goal: scalable = handles more load/features without a rewrite; low-maintenance = easy to change safely (small, well-separated pieces); robust = behaves correctly under edge cases and failures. Good relationships and clear class boundaries (the rest of these notes) are exactly the tools that buy you these.

3.The Three Types of LLD Interviews

Theoretical rounds are common at older, large tech companies and banks (e.g., Oracle). They test knowledge with theoretical questions about OOP and design patterns. Expect 30–45 minutes, three or four questions, and no real code (toy code at most). Design rounds appear at big MNCs like Amazon, PayPal, and Walmart. They're problem-solving, case-study rounds: roughly 45 minutes, one question plus discussion, where you write code for a few classes but aren't expected to make an end-to-end program run. Machine Coding rounds are the very first round at companies like Flipkart, Myntra, and Razorpay, and at essentially every startup. They want people who can actually implement: you write end-to-end working code from a set of requirements, usually over about 2.5 hours.

+ These aren't mutually exclusive across a single loop — a company can run a Design round and a Machine Coding round on the same day. Treat the three as modes to be ready for, not a single label per company.

4.Prerequisites for LLD

The foundation is programming experience in an OOP language — Java, C++, Python, or similar. On top of that, a good understanding of the OOP concepts: encapsulation, inheritance, abstraction, polymorphism, and the relationship trio of association, aggregation, and composition. Beyond the concepts, a good grasp of object-oriented design principles, especially SOLID. Finally, command of design patterns — a bonus for entry-level developers and effectively expected of senior ones.

• Covered here: the four pillars (recap) and the three has-a relationships + cardinality.
• Covered later: SOLID principles and design patterns.

+ One-line preview of SOLID so the term isn't a black box: Single responsibility, Open/closed, Liskov substitution, Interface segregation, Dependency inversion. + Design patterns are reusable, named solutions to recurring design problems (e.g., Factory, Strategy, Observer, Singleton). Both are the natural next layer on top of everything in these notes.

5.The Four Pillars of OOP (Recap)

Encapsulation bundles data and the methods that operate on it inside one unit and hides the internal fields, exposing controlled access. Its goals: reduce complexity and provide data security. Abstraction exposes what something does while hiding how it does it; goals: hide complexity and isolate the impact of changes (callers depend on a stable surface, not internals). Inheritance lets a class reuse another class's members through an "is-a" relationship; goals: eliminate redundant code and enable reuse. Polymorphism lets one object take many forms — the same call behaves differently depending on the actual type.

+ Polymorphism in Java has two concrete forms worth naming: compile-time (method overloading — same name, different parameters) and runtime (method overriding — subclass redefines a parent method, resolved by the actual object type via dynamic dispatch). Interviewers frequently ask for this distinction. See code §III · OOP pillars.

6.How to Represent an LLD

You could communicate a design with words — email, a meeting, a Slack message — but words breed misunderstanding ("I didn't mean that") and struggle to convey complex scenarios. Diagrams are better: a picture is worth a thousand words, they carry less ambiguity, and they make complex structures and scenarios easy to visualize. The remaining problem: if everyone draws their own way, diagrams stop being a shared language. UML (Unified Modelling Language) solves that by providing a standard for representing software concepts in diagrams.

+ UML defines many diagram types — class, sequence, use-case, activity, state, etc. For LLD the class diagram (static structure) and the sequence diagram (how objects interact over time) are the two you'll actually use most.

7.The UML Class Diagram

The class diagram is the workhorse of LLD: it shows the static structure — the classes and how they relate. Each class is a rectangle divided into three stacked compartments. The top holds the class name. The middle lists attributes (fields) as visibility name : Type. The bottom lists methods as visibility name(params) : ReturnType. Visibility markers are a small, fixed vocabulary.

┌──────────────────────────────┐
│           BankAccount        │
├──────────────────────────────┤
│ - accountNo : String         │
│ - balance   : double         │
├──────────────────────────────┤
│ + deposit(amount:double):void│
│ + getBalance() : double      │
└──────────────────────────────┘

+ Two more conventions worth memorizing: underlined members are static, and italic class or method names are abstract. + The attribute/method ordering convention is visibility name : type — note the type comes after the name with a colon, which is the reverse of Java's own int id; syntax. This trips people up.

8.Class Relationships — the "Has-A" Family

Take a simple requirement set for a Manager at "XYZ Limited": the manager is an employee of XYZ; the manager uses a swipe card to enter the premises; the manager has workers under them; the manager is responsible for a project's success; and the manager's salary is judged by that project's success. Each clause hides a different kind of relationship between objects, and naming them precisely is the whole game.

The deciding questions are always two: Is there ownership? and What happens to the child's lifecycle when the parent goes away? Crucially, aggregation and composition are special cases of association — they're association plus increasing strictness — which is exactly why it can be hard to decide which to use.

          ASSOCIATION  (no ownership, independent lives)
                │
                ▼ add an owner (loose)
          AGGREGATION  (owner, but child outlives parent)
                │
                ▼ tighten: child can't live without parent
          COMPOSITION  (child dies with parent)

+ In code these typically look like: association/aggregation = the parent holds a reference to an object created elsewhere and passed in; composition = the parent creates and owns the child internally so nothing outside can hold it. The lifecycle rule falls straight out of who constructs the object.

9.Association

Association is the loosest relationship. The objects use one another but each has its own lifetime and can exist without the other — there is no single owner. In the manager example, the manager and the swipe card use each other, yet each exists independently; neither owns the other. A classic textbook example: a teacher and a student — multiple students can have a single teacher and a single student can have multiple teachers. Neither creates or destroys the other.

Real-world examples: Doctor ↔ Patient, Author ↔ Reader, Driver ↔ Passenger on a single ride. + Direction matters: association can be unidirectional (A knows B, B doesn't know A) or bidirectional (both reference each other). In UML it's a plain solid line; an arrowhead indicates navigability/direction.

10.Aggregation

Aggregation is like association but with one owner. From the manager requirements, "the manager has workers who work under him" is aggregation: the Manager object owns the Worker objects, and a given Worker can't simultaneously belong to some other unrelated parent (a Worker doesn't belong to a SwipeCard). But — and this is the defining trait — a Worker has their own lifetime, disconnected from the Manager. If the Manager object is deleted, the Worker object does not die. Another clean example: a car may have passengers; they come and go, and they clearly outlive any single car.

Real-world examples: Department ◇── Professor (close the department, professors still exist and move elsewhere); Team ◇── Player; Playlist ◇── Song (delete the playlist, the songs remain). + UML notation: a hollow (unfilled) diamond on the owner/parent end of the line. Mnemonic: hollow diamond = loose hold.

11.Composition

Composition is the strictest has-a. The child object does not have an independent lifecycle, and if the parent is deleted all children go with it. One framing: "the manager and the project are dependent on each other and share the same lifetime" — the project won't succeed without a good manager and the manager's increments depend on the project — making them heavily interdependent, where if one is garbage-collected the other effectively is too. A cleaner, more standard example: a hotel has rooms — destroy the hotel and the rooms are destroyed with it. The rooms have no meaning or existence apart from the hotel.

+ A note on the source's primary example: framing Manager ↔ Project as composition is debatable — it's really describing a tight two-way dependency, not a contained-part-with-no-independent-existence. The textbook-clean composition test is the Hotel/Room or House/Room case, where the part literally cannot exist without the whole. Lean on those examples in an interview.

Real-world examples: House ◆── Room; Order ◆── OrderLineItem; Human ◆── Heart; Document ◆── Paragraph. + UML notation: a filled (solid) diamond on the parent end. Mnemonic: filled diamond = lifelong, fully owned. In code the parent usually news the child inside itself.

Putting the has-a family together

Deciding between the three can be genuinely tricky because aggregation and composition are subsets of association. Use this decision flow:

Q1: Is there a clear owner (parent has-a child)?
      NO  → ASSOCIATION
      YES → Q2
Q2: If the parent is deleted, must the child be deleted too?
      YES → COMPOSITION   (◆ filled diamond, parent creates child)
      NO  → AGGREGATION   (◇ hollow diamond, child passed in)

12.Cardinality

Cardinality is the count side of a relationship: for a link between two classes, how many objects participate at each end. You determine it by asking the question in both directions and combining the answers. The four canonical cases, worked through with a User class:

Read each as "User → X". 1:1 a user has exactly one passport and a passport belongs to one user. 1:M a user has many credit cards but each card belongs to one user. M:1 a user joins one school but a school holds many users. M:M a user enrolls in many courses and a course holds many users.

+ Standard UML multiplicity notation written at each line end: 1 (exactly one), 0..1 (optional one), * or 0..* (zero or more), 1..* (one or more). So a User-to-Course many-to-many is drawn User 1..* ──── 0..* Course (or *—*). + M:M is the case to watch — it almost always needs an extra class to hold the relationship plus any per-link data (enrollment date, grade). See Code Sample: Cardinality / Enrollment.

Section III — Code Samples

§OOP Pillars (encapsulation, inheritance, polymorphism, abstraction)

Links back to Notes §5. One file demonstrating all four pillars, including both overloading (compile-time) and overriding (runtime) polymorphism.

// Abstraction: callers use Shape without knowing each subclass's math.
abstract class Shape {
    abstract double area();              // "what", not "how"
}

// Inheritance: Circle IS-A Shape -> reuses the Shape type.
class Circle extends Shape {
    private double r;                    // Encapsulation: field is private
    Circle(double r) { this.r = r; }
    double area() { return Math.PI * r * r; }   // Polymorphism: overrides area()
}

class Rectangle extends Shape {
    private double w, h;
    Rectangle(double w, double h) { this.w = w; this.h = h; }
    double area() { return w * h; }             // runtime polymorphism (override)
}

class MathUtil {
    // Compile-time polymorphism: OVERLOADING (same name, diff params)
    static int max(int a, int b)       { return a > b ? a : b; }
    static double max(double a, double b) { return a > b ? a : b; }
}

public class OopPillars {
    public static void main(String[] args) {
        Shape[] shapes = { new Circle(2), new Rectangle(3, 4) };
        for (Shape s : shapes) // Same call s.area() -> different behavior per actual type
        System.out.printf("%s area = %.2f%n", s.getClass().getSimpleName(), s.area());

        System.out.println("max(int)    = " + MathUtil.max(7, 3));
        System.out.println("max(double) = " + MathUtil.max(7.5, 9.1));
    }
}

Circle area = 12.57
Rectangle area = 12.00
max(int)    = 7
max(double) = 9.1

§A UML Class as Java (visibility markers)

Links back to Notes §7. The three-compartment diagram (name / attributes / methods) realized in code, showing how UML visibility maps to Java keywords.

public class BankAccount {
    // ---- UML middle compartment: attributes ----
    private String accountNo;   // UML  - accountNo : String   (-, private)
    private double balance;     // UML  - balance   : double

    public BankAccount(String accountNo) {  // UML + (public) constructor
        this.accountNo = accountNo;
        this.balance = 0.0;
    }

    // ---- UML bottom compartment: methods ----
    public void deposit(double amount) {        // + deposit(amount:double):void
        if (amount <= 0) throw new IllegalArgumentException("amount must be > 0");
        balance += amount;
    }

    public double getBalance() { return balance; }   // + getBalance() : double

    protected String maskedNo() {                    // # (protected)
        return "****" + accountNo.substring(accountNo.length() - 2);
    }

    public static void main(String[] args) {
        BankAccount acc = new BankAccount("100042");
        acc.deposit(500);
        acc.deposit(250);
        System.out.println("Account " + acc.maskedNo() + " balance = " + acc.getBalance());
    }
}

Account ****42 balance = 750.0

§Association — Teacher / Student

Links back to Notes §9. No ownership; both objects are created independently and merely reference each other.

import java.util.*;

class Teacher {
    String name;
    Teacher(String name) { this.name = name; }
}

class Student {
    String name;
    // Association: a student references teachers but does NOT own them.
    List<Teacher> teachers = new ArrayList<>();
    Student(String name) { this.name = name; }
    void addTeacher(Teacher t) { teachers.add(t); }  // many-to-many link
}

public class AssociationDemo {
    public static void main(String[] args) {
        Teacher alice = new Teacher("Alice");   // created independently
        Teacher bob   = new Teacher("Bob");
        Student sam = new Student("Sam");
        sam.addTeacher(alice);
        sam.addTeacher(bob);

        System.out.print(sam.name + " is taught by: ");
        for (Teacher t : sam.teachers) System.out.print(t.name + " ");
        System.out.println();
        // Deleting Sam (drop reference) does NOT affect the teachers:
        sam = null;
        System.out.println("After Sam is gone, Alice still exists: " + alice.name);
    }
}

Sam is taught by: Alice Bob
After Sam is gone, Alice still exists: Alice

§Aggregation — Department / Professor

Links back to Notes §10. Ownership exists (Department has-a Professor), but professors are created outside and survive the department.

import java.util.*;

class Professor {
    String name;
    Professor(String name) { this.name = name; }
}

class Department {
    String name;
    // Aggregation: professors are PASSED IN, not created here.
    private List<Professor> professors;
    Department(String name, List<Professor> professors) {
        this.name = name;
        this.professors = professors;   // shares externally-owned objects
    }
    List<Professor> getProfessors() { return professors; }
}

public class AggregationDemo {
    public static void main(String[] args) {
        // Professors exist BEFORE the department and can outlive it.
        Professor p1 = new Professor("Dr. Rao");
        Professor p2 = new Professor("Dr. Lee");
        List<Professor> staff = new ArrayList<>(List.of(p1, p2));

        Department cs = new Department("Computer Science", staff);
        System.out.println(cs.name + " has " + cs.getProfessors().size() + " profs");

        cs = null;  // "delete" the department
        System.out.println("Dept gone, but " + p1.name + " still exists.");
    }
}

Computer Science has 2 profs
Dept gone, but Dr. Rao still exists.

§Composition — House / Room

Links back to Notes §11. The House creates and owns its Rooms internally; nothing outside can hold them, so they die with the House.

import java.util.*;

public class CompositionDemo {

    // Room is an inner/owned part. It has no meaning without a House.
    static class House {
        private final List<Room> rooms = new ArrayList<>();

        House(int roomCount) {
            // Composition: the House CREATES its own rooms internally.
            for (int i = 1; i <= roomCount; i++) rooms.add(new Room("Room-" + i));
        }
        int roomCount() { return rooms.size(); }

        // Rooms are never exposed for outside ownership.
        class Room {
            private final String label;
            Room(String label) { this.label = label; }
        }
    }

    public static void main(String[] args) {
        House h = new House(3);
        System.out.println("House has " + h.roomCount() + " rooms.");

        h = null;  // delete the house
        // No external reference to any Room exists, so all rooms are
        // unreachable now -> eligible for garbage collection with the house.
        System.out.println("House deleted -> its rooms are gone too.");
    }
}

House has 3 rooms.
House deleted -> its rooms are gone too.

§Cardinality — the four cases + an M:M join class

Links back to Notes §12. Shows 1:1, 1:M, M:1 as fields/lists and the M:M case modeled with an explicit Enrollment link.

import java.util.*;

class Passport { String number; Passport(String n){number=n;} }      // 1:1
class CreditCard { String last4; CreditCard(String n){last4=n;} }     // 1:M
class School { String name; School(String n){name=n;} }               // M:1
class Course { String title; Course(String t){title=t;} }             // M:M

class User {
    String name;
    Passport passport;                 // 1:1  user has exactly one
    List<CreditCard> cards = new ArrayList<>();  // 1:M user has many
    School school;                     // M:1  many users -> one school
    User(String name){ this.name = name; }
}

// M:M needs a join class to link the two sides (+ per-link data).
class Enrollment {
    User user; Course course; String date;
    Enrollment(User u, Course c, String d){ user=u; course=c; date=d; }
}

public class CardinalityDemo {
    public static void main(String[] args) {
        User u = new User("Sam");
        u.passport = new Passport("P123");           // 1:1
        u.cards.add(new CreditCard("4242"));         // 1:M
        u.cards.add(new CreditCard("1881"));
        u.school = new School("MIT");                 // M:1

        Course algo = new Course("Algorithms");      // M:M via Enrollment
        List<Enrollment> enrollments = new ArrayList<>();
        enrollments.add(new Enrollment(u, algo, "2026-06-23"));

        System.out.println("1:1  passport      = " + u.passport.number);
        System.out.println("1:M  #cards        = " + u.cards.size());
        System.out.println("M:1  school        = " + u.school.name);
        System.out.println("M:M  enrolled in   = " + enrollments.get(0).course.title);
    }
}

1:1  passport      = P123
1:M  #cards        = 2
M:1  school        = MIT
M:M  enrolled in   = Algorithms

Section IV — Hands-On Lab

# 1. Check the toolchain (need JDK 17+; works on 8+)
java -version
javac -version

# 2. Make a working folder
mkdir lld-lab && cd lld-lab

# 3. Compile + run pattern used in every exercise
#    (put each public class in a file matching its name)
javac Solution.java
java Solution

# Tip: one file can hold many classes; only ONE public class,
# and the file name must match that public class.

public class Profile {
    private String email;       // UML:  -  (private)
    protected int id;           // UML:  #  (protected)
    String nickname;            // UML:  ~  (package-private, no modifier)
    public String getEmail() {  // UML:  +  (public)
        return email;
    }
    public Profile(String email, int id, String nickname) {
        this.email = email; this.id = id; this.nickname = nickname;
    }
    public static void main(String[] args) {
        Profile p = new Profile("sam@x.com", 1, "sammy");
        System.out.println(p.getEmail());
    }
}

sam@x.com

import java.util.*;

class Player { String name; Player(String n){ name = n; } }

class Team {
    String name;
    private List<Player> players;
    Team(String name, List<Player> players){  // players passed in -> aggregation
        this.name = name; this.players = players;
    }
}

public class TeamDemo {
    public static void main(String[] args) {
        Player p1 = new Player("Ada");
        Player p2 = new Player("Linus");
        Team t = new Team("Hackers", new ArrayList<>(List.of(p1, p2)));
        System.out.println("Team built: " + t.name);
        t = null;                                  // delete the team
        System.out.println("Team gone, player alive: " + p1.name);
    }
}

Team built: Hackers
Team gone, player alive: Ada

public class Computer {
    private final CPU cpu;     // owned parts, created internally
    private final RAM ram;

    Computer(String cpuModel, int gb) {
        this.cpu = new CPU(cpuModel);   // composition: parent NEWs the children
        this.ram = new RAM(gb);
    }
    String spec() { return cpu.model + " / " + ram.gb + "GB"; }

    private static class CPU { String model; CPU(String m){ model = m; } }
    private static class RAM { int gb;      RAM(int g){ gb = g; } }

    public static void main(String[] args) {
        Computer pc = new Computer("M3", 16);
        System.out.println("Spec: " + pc.spec());
        pc = null;  // both CPU and RAM become unreachable with it
        System.out.println("Computer deleted -> CPU & RAM gone too.");
    }
}

Spec: M3 / 16GB
Computer deleted -> CPU & RAM gone too.

import java.util.*;

class LibraryCard { String id; LibraryCard(String i){ id = i; } }
class Fine { double amt; Fine(double a){ amt = a; } }
class Branch { String name; Branch(String n){ name = n; } }
class Book { String title; Book(String t){ title = t; } }

class Member {
    String name;
    LibraryCard card;                       // 1:1
    List<Fine> fines = new ArrayList<>();    // 1:M
    Branch branch;                          // M:1
    Member(String n){ name = n; }
}

class Loan {                                // M:M join (+ link data)
    Member member; Book book; String due;
    Loan(Member m, Book b, String d){ member=m; book=b; due=d; }
}

public class LibrarySystem {
    public static void main(String[] args) {
        Member m = new Member("Sam");
        m.card = new LibraryCard("LC-1");            // 1:1
        m.fines.add(new Fine(2.50));                 // 1:M
        m.fines.add(new Fine(1.00));
        m.branch = new Branch("Downtown");           // M:1

        Book b = new Book("Clean Code");             // M:M via Loan
        List<Loan> loans = new ArrayList<>();
        loans.add(new Loan(m, b, "2026-07-01"));

        System.out.println("1:1 card   = " + m.card.id);
        System.out.println("1:M fines  = " + m.fines.size());
        System.out.println("M:1 branch = " + m.branch.name);
        System.out.println("M:M loan   = " + loans.get(0).book.title);
    }
}

1:1 card   = LC-1
1:M fines  = 2
M:1 branch = Downtown
M:M loan   = Clean Code

import java.util.*;

// 1:1  Vehicle --- RegistrationPlate
class RegistrationPlate {
    private final String number;
    RegistrationPlate(String number){ this.number = number; }
    String number(){ return number; }
}

class Vehicle {
    private final RegistrationPlate plate;   // 1:1 composition-ish (plate owned)
    Vehicle(String plateNo){ this.plate = new RegistrationPlate(plateNo); }
    String plate(){ return plate.number(); }
}

class ParkingSpot {
    private final int id;
    private Vehicle vehicle;                  // AGGREGATION, 0..1 (vehicle passed in)
    ParkingSpot(int id){ this.id = id; }
    boolean isFree(){ return vehicle == null; }
    boolean park(Vehicle v){                  // receive, never create -> aggregation
        if (!isFree()) return false;
        this.vehicle = v;
        return true;
    }
    void leave(){ this.vehicle = null; }      // vehicle survives; just unlinked
    int id(){ return id; }
}

public class ParkingLot {
    // COMPOSITION, 1:M  ParkingLot <>--- ParkingSpot (spots created inside)
    private final List<ParkingSpot> spots = new ArrayList<>();

    ParkingLot(int size){
        for (int i = 1; i <= size; i++) spots.add(new ParkingSpot(i)); // lot OWNS spots
    }
    ParkingSpot park(Vehicle v){
        for (ParkingSpot s : spots) if (s.park(v)) return s;
        return null;  // lot full
    }
    long availableSpots(){
        return spots.stream().filter(ParkingSpot::isFree).count();
    }

    public static void main(String[] args) {
        ParkingLot lot = new ParkingLot(3);
        Vehicle car1 = new Vehicle("KA-01-1111");   // vehicles exist independently
        Vehicle car2 = new Vehicle("KA-02-2222");

        ParkingSpot s1 = lot.park(car1);
        lot.park(car2);
        System.out.println("Parked " + car1.plate() + " at spot " + s1.id());
        System.out.println("Available: " + lot.availableSpots());

        s1.leave();   // car1 drives off; the Vehicle object still exists
        System.out.println("After " + car1.plate() + " leaves -> Available: " + lot.availableSpots());
    }
}

Parked KA-01-1111 at spot 1
Available: 1
After KA-01-1111 leaves -> Available: 2

★Key Takeaways

• LLD = blueprint before code. Detailed planning first beats jumping in or staying high-level — clear roadmap, aligned team, fewer errors.
• Three interview formats: Theoretical (theory, no code), Design (a few classes + discussion), Machine Coding (full working code). Know which you're facing.
• Four OOP pillars by goal: Encapsulation (reduce complexity + security), Abstraction (hide complexity + isolate change), Inheritance (reuse), Polymorphism (one form, many behaviors).
• Represent designs with diagrams, standardized by UML. Words are ambiguous; the class diagram is the LLD workhorse — one box, three compartments, + - # ~ visibility.
• Has-a family is a spectrum: Association (no owner) ⊃ Aggregation (owner, child survives) ⊃ Composition (child dies with parent). UML: line / ◇ / ◆.
• The one test that decides it all: "If I delete the parent, must the child die?" Yes → composition; No-but-owned → aggregation; No-owner → association.
• Cardinality drives structure: 1:1 and M:1 → a field; 1:M → a list; M:M → a join class. Always state your read direction.