About datarekha
datarekha is a free, interactive platform for learning the modern data and AI stack. The goal is simple: teach the things you'll actually use in industry, in the order a new hire actually learns them — every code example shown with its real, verified output, every concept checked with a quiz.
It has grown into a deep, end-to-end curriculum: hundreds of lessons across many subjects, organized into clear categories — Foundation, Data & Stats, ML, AI, and more — plus a full GATE DA exam-prep section, guided learning tracks, and a long-form blog — all built to a single bar, that the idea should genuinely click.
What's inside
- Foundations & data stack — Python (from syntax to async LLM apps, Pydantic, FastAPI), NumPy, Pandas, Matplotlib & Seaborn.
- SQL & PySpark — analytics-grade SQL through window functions and recursive CTEs, warehouse dialects, and Spark internals end to end.
- Business Analytics — unit economics, P&L, RFM, funnels and cohorts, decision analysis, forecasting, A/B testing, and storytelling with data.
- Math, ML & Deep Learning — the math an ML engineer really uses, scikit-learn, XGBoost, and PyTorch-first deep learning with transformers, fine-tuning, and generative models.
- MLOps — Docker, CI/CD, MLflow, serving, drift monitoring, and incident response.
- Generative & Agentic AI — LLMs in practice, RAG, evals, plus the agent-interoperability stack (MCP, A2A, ACP, ANP), LangChain, LangGraph, the Microsoft Agent Framework, and Google ADK.
- Data Structures & Algorithms — the algorithmic core for data and AI, including the probabilistic structures behind real dedup and retrieval.
- GATE DA — an exam-grade track for GATE Data Science & Artificial Intelligence: probability, statistics, linear algebra, calculus, Python, DSA, DBMS, warehousing, machine learning, and AI search & reasoning, with solved previous-year questions.
How you learn here
- Every output is real. Code samples ship with the exact output they produce — run and verified before publishing, never guessed. Copy any snippet and it behaves precisely as shown.
- See it move. The hard ideas are taught with 100+ hand-built interactive explorables and animations, not static diagrams.
- Check yourself. A quiz on every concept, and curated guided tracks that sequence the curriculum end-to-end with sensible pacing.
The blog
Alongside the lessons, datarekha publishes long-form essays on production AI, GATE DA, and career growth — including a data-backed series on developer wellbeing and the modern workplace.
Who it's for
Students, career-switchers, and working data, ML, and AI professionals who want real depth without the fluff — whether you're shipping models, prepping for GATE, or just trying to understand a hard idea properly.
Who makes this, and our standards
datarekha is written and maintained by a practitioner with hands-on industry experience across data, machine learning, and AI engineering — not auto-generated filler. Every lesson is built to be technically correct and genuinely useful:
- Accuracy first. Concepts are checked against primary sources — official documentation, papers, and language/library specs — and lessons carry an "updated" date so you can see they're maintained.
- We fix what's wrong. Spotted an error? Email us and we correct it — see Contact. Corrections are part of how the curriculum stays reliable.
- Kept current. The data and AI stack moves fast, so lessons are revised as tools, versions, and best practices change.
How it's built
A static site generated with Astro, styled with Tailwind CSS v4, with content authored in MDX. Interactive widgets are islands that hydrate only when needed, and hovered links are prerendered, so pages open close to instantly. Python and SQL run in your browser and load lazily, so the landing pages stay fast.
Get in touch
Questions, corrections, or ideas? Email the team at team@datarekha.com.