Python
Code · Data
Write small programs to transform data, call APIs and automate repeatable work.
All capabilities · Programming foundations
Write production-quality Python for AI work
Write clean, tested Python with virtualenvs, typing, packaging and async basics — the lingua franca of every AI role.
Explore 4 tools for this projectMarket relevance
Which roles ask for this — and how often
Share of job postings in India, per role, that name this capability.
AI Quality Assurance / GenAI Test Engineer100%Computer Vision Engineer89%Machine Learning Engineer89%AI Automation Specialist80%AI Engineer78%AI Security Engineer78%AI Technical Trainer / Instructor76%AI-enabled Data Analyst70%Forward Deployed Engineer (AI)68%Prompt Engineer / AI Workflow Specialist66%AI Support Engineer46%AI Solutions / Pre-sales Engineer30%AI Search Optimization Specialist25%AI Data Annotator / Labeling QA15%
What employers mean
You should be able to…
- Write clean, typed Python functions instead of untyped notebook cells
- Set up a virtualenv/uv project with pinned dependencies (requirements.txt or pyproject.toml)
- Write pytest unit tests with fixtures and mocks for functions that call external APIs
- Use async/await to call multiple LLM or third-party APIs concurrently
- Package a script into an importable module with a proper __init__.py and CLI entrypoint
- Handle exceptions and retries around flaky network calls instead of letting scripts crash
- Read and refactor someone else's Python codebase without breaking existing tests
Learn — free, link-checked
The few resources that matter
Read · beginner · 20 min · docs.pytest.org
pytest — Get Started
pytest is the de facto testing tool employers expect; this gets you writing and running real tests in minutes. — pytest
Read · beginner · 90 min · docs.python.org
The Python Tutorial
Ground yourself in modern Python syntax and idioms straight from the source before layering AI-specific libraries on top. — Python Software Foundation
Read · beginner · 120 min · fastapi.tiangolo.com
FastAPI — User Guide / Tutorial
The official, example-driven walkthrough from hello-world to a typed, validated, auto-documented API — the framework Indian AI job posts name most. — FastAPI (tiangolo)
Watch · beginner · 360 min · youtube.com
Data Structures and Algorithms in Python - Full Course for Beginners
Builds every core data structure from scratch in Python, cementing both DSA concepts and idiomatic Python at once. — freeCodeCamp.org
Read · intermediate · 25 min · docs.github.com
Building and testing Python
Shows exactly how to wire pytest and matrix Python versions into a CI workflow — the setup interviewers ask you to describe. — GitHub Docs
Read · intermediate · 30 min · docs.python.org
typing — Support for type hints
Type hints are expected in production AI codebases (Pydantic, FastAPI); the canonical reference for annotating functions and generics correctly. — Python Software Foundation
Read · intermediate · 35 min · docs.python.org
asyncio — Asynchronous I/O
LLM API calls are I/O-bound; understanding async/await lets you fan out concurrent model calls instead of blocking one at a time. — Python Software Foundation
Tools for practice
Choose a tool for the job
Start with one tool for each part of your project. You don’t need to learn them all.
4 tools to explore
uv
Code
Create isolated Python environments and keep project dependencies reproducible.
pytest
Test
Turn expected behaviour and failure cases into repeatable Python tests.
Ruff
Code
Catch common Python mistakes and apply consistent formatting before review.
Practice
Support-ticket triage CLI as a typed, tested Python package
Build a Python package that reads a folder of labelled customer-support tickets, classifies each into a category (refund, fraud, technical) with rules plus one small model call, and writes a summary CSV. Wrap it in a Typer CLI, add asyncio batch processing so a folder of 500 tickets finishes in one pass, and cover the classifier and I/O with pytest fixtures. Type-hint everything and keep mypy at zero errors as the package grows.
Start from
Milestones
- Slice the dataset into per-ticket JSON and load it into typed records · ~5.5h
- Write the rule + model-call classifier behind one interface, with pytest fixtures · ~7.5h
- Wrap it in a Typer CLI and add asyncio batching over the 500-ticket folder · ~7.5h
- Drive mypy to zero errors and coverage past 80% · ~4.5h
- Write the README with sample input/output and the timing numbers · ~3h
Done when
- Runs via `python -m triage --input tickets/ --output summary.csv` with no crashes on malformed input
- At least 80% pytest coverage on the classification and I/O modules, run via `pytest --cov`
- Uses type hints throughout and passes `mypy` with zero errors
- Processes 500 mock tickets concurrently with asyncio in under 10 seconds on a laptop
Prove it
Evidence a recruiter can check
- The `pytest --cov` table pasted in the README, showing 80%+ on the classification and I/O modules
- A clean `mypy` run over the whole package, including the annotations you added for untyped third-party calls
- Serial vs asyncio wall-clock numbers for the same 500 tickets, measured on your machine and stated in the README
- The malformed-ticket test you wrote first, and the commit that made the CLI survive it instead of crashing
Signal it
Built a typed, async Python CLI that classifies 500 support tickets in a single pass — 80%+ pytest coverage on the classifier and a zero-error mypy run across the package.
Interview
Questions you'll get asked
- What's the difference between a list and a generator, and when would you use each?
- Explain GIL — does multithreading help a CPU-bound vs an I/O-bound Python task?
- How would you structure a Python project that calls an LLM API with retries and rate limiting?
- What does `async def` actually buy you over a regular function when calling 5 APIs?
- How do you mock an external API call in a pytest test?
- What's the difference between `@staticmethod`, `@classmethod` and a regular method?
- Walk me through how you'd add type hints to a function that returns either a dict or None.