Choosing a Language

What a programming language is

A programming language is a precise, human-readable notation for instructions that will ultimately run on a CPU (Level 0 refresher). "Precise" is the point: English tolerates ambiguity, computers don't. Every language is a different trade-off between easy for humans to write and easy for machines to run fast.

This roadmap teaches three — Python, Java and C++ — because between them they cover the whole trade-off spectrum, dominate industry and interviews, and once you know these three, every other language is a weekend of adjustment.

The same program, three ways

Print the squares of 1–5, skipping 3:

# Python
for i in range(1, 6):
    if i == 3:
        continue
    print(i * i)

// Java
public class Squares {
    public static void main(String[] args) {
        for (int i = 1; i <= 5; i++) {
            if (i == 3) continue;
            System.out.println(i * i);
        }
    }
}

// C++
#include <iostream>

int main() {
    for (int i = 1; i <= 5; i++) {
        if (i == 3) continue;
        std::cout << i * i << "\n";
    }
    return 0;
}

Same logic, same structure — different ceremony. That observation should be encouraging: you are learning programming once, in three accents.

Python — the language that reads like pseudocode

• Born: 1991, Guido van Rossum, Netherlands. Design goal: code should be readable — whitespace and plain words over braces and symbols.
• How it runs: an interpreter executes your file directly — no separate compile step. Type, run, see results. (It's also dynamically typed — more on that next page.)
• Strengths: fastest language to write and learn; enormous libraries — it is the language of AI/ML (Level 11), data science and scripting.
• Weaknesses: slow to execute relative to Java/C++ (often 10–100× for pure computation); dynamic typing lets some bugs hide until runtime.
• Industry: Google (massive Python user), Instagram (backend is Django, a Python framework), Netflix, every AI lab; the default glue language of the industry.
• Interviews: the most popular interview language — concise code means more thinking time. All DSA solutions in Levels 2–4 lead with Python.

Java — the language of big systems

• Born: 1995, James Gosling at Sun Microsystems. Pitch: "write once, run anywhere" — Java compiles to bytecode that runs on the JVM (Java Virtual Machine), a program that makes all operating systems look identical to your code.
• How it runs: compile → bytecode → JVM executes it (with heavy optimization at runtime). Statically typed: every variable's type is declared and checked before the program runs.
• Strengths: types catch bugs early; outstanding tooling; the JVM is battle-tested over 30 years; threads and concurrency are first-class. Built for codebases with hundreds of engineers.
• Weaknesses: verbose (see the ceremony above); slower iteration than Python; "enterprise" reputation earned honestly.
• Industry: Amazon's backend is famously Java-heavy; LinkedIn, Uber, Netflix's services, most banks; Android apps (via Kotlin, Java's modern cousin); Spring Boot (Level 7) is the dominant enterprise backend framework.
• Interviews: extremely common; for LLD/machine-coding rounds (Level 5), Java's explicit interfaces and access modifiers actually help you demonstrate design.

C++ — the language closest to the metal

• Born: 1985, Bjarne Stroustrup at Bell Labs — C (the 1972 systems language) extended with object-oriented features. "C with classes."
• How it runs: a compiler translates your code straight to machine code for a specific CPU — no interpreter, no virtual machine, nothing between you and the hardware. Statically typed.
• Strengths: the fastest of the three, with manual control of memory — you decide exactly what is allocated and freed. When microseconds or megabytes matter, C++ (or its young rival Rust) is the answer.
• Weaknesses: that control is work and risk — memory bugs (crashes, security holes) that Java/Python make impossible are everyday hazards in C++; the language is huge; the learning curve is real.
• Industry: game engines (Unreal), browsers (Chrome is C++), trading systems where nanoseconds are money, operating systems, embedded devices, the internals of Python and the JVM themselves.
• Interviews: the traditional choice of competitive programmers (fastest execution under tight limits); some infrastructure teams interview in it.

Side-by-side summary

Common beginner mistakes

• Language-hopping. Two weeks of Python, then "should I switch to JavaScript?", then Rust… Pick one, reach "can build a small project," then branch. Depth transfers; dabbling doesn't.
• Believing "X is dead." Python (1991), Java (1995) and C++ (1985) have outlived every obituary. Companies run on decades-old code that needs decades more of engineers.
• Confusing "easy to start" with "less professional." Python's simplicity runs Instagram and trains GPT-class models. Simple ≠ toy.
• Thinking speed of the language is speed of the product. Most products are limited by network and database time (Level 0), not CPU. Engineer time is usually the scarcer resource — which is exactly why "slow" Python thrives.

Interview perspective

Practice

1. Setup: install Python (python.org), then run print("hello, world") — first from the interactive prompt, then from a file. (Java and C++ setups come when you need them; an online editor like replit.com works for all three today.)
2. Modify the squares program above to print cubes of 1–10, skipping multiples of 4 — in Python first, then port it to one other language and notice exactly what changed.
3. In one sentence each: which of the three languages would you pick to (a) train an ML model, (b) build a bank's transaction backend, (c) write a game engine's physics loop — and why?

Next: Variables & Data Types — your first real building blocks.