Data Science Katas

A hands-on collection of 250+ exercises, cheatsheets, and interview prep materials covering the core skills tested in Data Science and ML Engineering interviews.

Every section follows a question / solution format so you can practice actively, not just read passively.

If you find this useful, a star helps others discover it and keeps me motivated.

---

What's Inside

Section	Exercises	Format	Topics
Python OOP	40+	Notebooks + .py	Classes, inheritance, encapsulation, polymorphism, dunder methods, abstract classes
SQL	30 sets	Notebooks + PostgreSQL	Joins, CTEs, window functions, subqueries + 22 real-world applied scenarios
Machine Learning	10+ notebooks, 28 cheatsheets	Notebooks + Markdown	EDA, regression, SVMs, trees, ensembles, PCA, metrics, feature engineering
PyTorch	2 tutorials + cheatsheet	Notebooks	Tensors, model building, training loops, neural net fundamentals
LangGraph	10 exercises	Notebooks	Graph construction, conditional/looping flows, ReAct agents, RAG agents
A/B Testing	5 notebooks + 3 cheatsheets	Notebooks + Markdown	Frequentist tests, Bayesian testing, multivariant tests, interview Q&A
Algorithms & Data Structures	10 implementations + theory	.py + Markdown	Search, sorting, arrays, linked lists, hash tables, heaps, Big-O

---

Python OOP

Practice object-oriented Python from first principles through advanced patterns.

python/
├── classes 101/         # 8 notebook pairs: instances, class vars, methods,
│                        #   inheritance, dunder methods, property decorators,
│                        #   abstract classes
├── encapsulation/       # 6 exercise pairs
├── polymorphism/        # 2 notebook pairs + 10 extra .py exercises
└── extra_inheritance/   # 8 .py exercise pairs

SQL

8 foundational topics plus 22 real-world applied scenarios, all runnable against a local PostgreSQL container.

sql/
├── basic_exercises/     # CREATE/INSERT, SELECT/GROUP BY, UNION, JOINs,
│                        #   CTEs, subqueries, window functions
├── applied_exercises/   # 22 scenarios: events, trades, viewership,
│                        #   transactions, sessions, reviews, signups...
├── migrations/          # DDL/DML scripts for each applied exercise
├── theory_questions.ipynb
└── migrate.py           # One command to set up all tables

Setup: requires Docker. See SQL setup instructions below.

Machine Learning

Cheatsheets and hands-on notebooks spanning the full ML interview syllabus.

machine_learning/
├── 00_eda/              # Univariate, bivariate, multivariate EDA guides
├── 01_regression/       # Linear, logistic, polynomial (notebooks + cheatsheets + QA)
├── 02_SVMs/             # Exercise, cheatsheet, QA
├── 03_decision_trees/   # Cheatsheet + QA
├── 04_ensemble_learning_random_forests/
├── 05_dimensionality_reduction/
├── 06_metrics/          # Regression + classification metrics
└── 07_feature_engineering/  # 9 cheatsheets: missing data, encoding, scaling,
                             #   outliers, feature selection, class imbalance,
                             #   splitting, pipelines, regularization

PyTorch

Neural network fundamentals and hands-on PyTorch workflows.

pytorch/
├── intro_nn_cheatsheet.md         # Neurons, perceptrons, activations, backprop
└── tutorials/
    ├── 00_pytorch_fundamentals.ipynb  # Tensors and operations
    └── 01_pytorch_workflow.ipynb      # Model building and training loops

LangGraph

Build agentic LLM applications from simple graphs to full RAG agents.

langgraph/
├── 01-05 graph exercises   # Sequential, conditional, looping graphs
└── agents/
    ├── 01_simple_bot        # Basic chatbot
    ├── 02_agent_bot         # Tool-using agent
    ├── 03_react_agent       # ReAct pattern
    ├── 04_agent_drafter     # Drafting agent
    └── 05_rag_agent         # Retrieval-augmented generation

A/B Testing

End-to-end experimentation: theory, statistical tests, and hands-on notebooks.

ab_testing/
├── AB_TESTING_CHEATSHEET.md          # Full workflow: hypotheses to decisions
├── AB_TESTING_GUIDELINE.md           # Step-by-step with PICOT criteria
├── ab_testing_statistical_tests.md   # Z-test, t-test, chi-square, Mann-Whitney,
│                                     #   bootstrap, Bayesian, ANOVA...
├── 01_ecommerce_conversion_rate      # Frequentist conversion test
├── 02_saas_session_duration           # Session duration test
├── 03_bayesian_ab_testing            # Bayesian approach
├── 04_multivariant_landing_page      # Multivariant test
└── 05_ab_testing_interview_qa        # Interview Q&A

Algorithms & Data Structures

Core implementations and theory notes for coding interviews.

algorithms_data_structures/
├── algorithms/          # linear search, binary search (iterative + recursive),
│                        #   quicksort, merge sort (array + linked list)
├── data_structures/     # arrays, linked lists, hash tables
└── NOTES.md             # Theory: stacks, queues, heaps, union find, Big-O

---

Getting Started

Prerequisites

• Python >= 3.10
• Poetry
• Docker (for SQL exercises only)

Installation

git clone https://github.com/amaldu/data-science-katas.git
cd data-science-katas
poetry install
poetry shell

Jupyter Kernel Setup

poetry run python -m ipykernel install --user --name=ds-katas --display-name "DS Katas"

Then select the DS Katas kernel when opening notebooks in VS Code or Jupyter.

Instructions on How to Run the SQL Exercises

1. Make sure Docker is running
2. Start the PostgreSQL container:

docker compose up -d

3. Connect using the PostgreSQL extension in VS Code (credentials are in docker-compose.yaml)
4. Run the migration script to create all tables:

poetry run python sql/migrate.py

5. Open any notebook in sql/ and add %%sql at the top of SQL cells

---

Contributing

Found an error or want to add exercises? Open an issue or submit a PR.

License

MIT