Question 1

What is Python's descriptor protocol and what built-in features use it?

Accepted Answer

Objects implementing __get__, __set__, and/or __delete__ that control attribute access — used internally to implement @property, @classmethod, and @staticmethod. When you access obj.attr, Python checks if the class's attr implements __get__. @property is a descriptor: reading the attribute calls the getter function. This mechanism powers most of Python's attribute access magic.

Question 2

What is a metaclass in Python?

Accepted Answer

A class whose instances are classes — used to customize class creation, e.g., ORMs use metaclasses to map class attributes to database columns. type is Python's default metaclass: type('MyClass', (Base,), {'attr': val}) creates a class. Custom metaclasses intercept class creation to add validation, register subclasses, or auto-generate methods. Django's Model uses a metaclass.

Question 3

What is uvloop and why does it significantly improve asyncio performance?

Accepted Answer

A fast event loop implementation for asyncio based on libuv (the same engine powering Node.js), 2–4x faster than the default asyncio event loop. The default asyncio event loop is pure Python. uvloop is written in Cython and wraps libuv — a battle-tested C event loop. Drop-in replacement: asyncio.set_event_loop_policy(uvloop.EventLoopPolicy()). Used in Uvicorn, the ASGI server behind FastAPI.

Question 4

For CPU-bound parallel work in Python, which executor should you use and why?

Accepted Answer

ProcessPoolExecutor — it creates separate processes that bypass the GIL, achieving true CPU parallelism. The GIL prevents multiple threads from running Python bytecode simultaneously. ProcessPoolExecutor spawns separate interpreter processes — each with its own GIL — achieving true parallelism. Use ThreadPoolExecutor only for I/O-bound tasks where threads release the GIL during I/O waits.

Question 5

What is __new__ in Python and how does it differ from __init__?

Accepted Answer

__new__ creates and returns the instance (called first); __init__ initializes it (called second) — __new__ is used to control object creation for immutable types and Singletons. For immutable types (int, str, tuple), you must use __new__ since __init__ can't change a value after allocation. For Singleton: override __new__ to return the existing instance if it already exists. __new__ receives the class; __init__ receives the created instance.

Python — Senior Dev Interview Questions

More Senior Dev Topics

Python at Other Levels