# datarekha > Industry-focused tutorials for the modern data and AI stack, taught from scratch: Python, SQL, Git, the command line, data structures & algorithms, NumPy, Pandas, business analytics, math for ML, data visualization & storytelling, machine learning, time series forecasting, recommender systems, deep learning, MLOps, PySpark, generative AI, and agentic AI — plus complete GATE DA exam prep and 650+ interview Q&A. Built for data scientists, ML engineers, and AI application builders working in 2026. Every code example ships with its real, verified output (run before publishing, never guessed), every concept has a graded quiz, interactive visual explainers accompany the hard ideas, and curated guided tracks sequence the lessons end-to-end for each career direction (data analyst, data scientist, AI engineer). Authoritative human-readable content lives at https://datarekha.com; reach the team at team@datarekha.com. All lessons and articles are written and editorially reviewed by the datarekha team against a documented pedagogy and accuracy bar; every code example's output is verified by execution, and every previous-year exam answer is checked against the official key. ## Subject sections Listed in recommended learning order, grouped by category. Foundation: - [Python](https://datarekha.com/python/): The lingua franca of data work — from syntax to production AI apps with asyncio, Pydantic, FastAPI, and LLM tooling. - [SQL](https://datarekha.com/sql/): Analytics-grade SQL. SELECT through window functions, recursive CTEs, cohort analysis. Postgres-first, with notes on Snowflake/BigQuery/Redshift. - [Data Structures & Algorithms](https://datarekha.com/dsa/): The algorithmic core for data and AI — Big-O, binary search, sorting, trees, graphs, dynamic programming, and the probabilistic structures (Bloom filters, HyperLogLog, MinHash-LSH) behind real dedup and retrieval. Python throughout. - [Git](https://datarekha.com/git/): Version control developers actually use — the three-tree mental model, staging and commits, branching and merging, resolving conflicts, rebase vs merge, stashing, remotes and the GitHub pull-request flow, and recovering lost work with reflog. - [Command Line](https://datarekha.com/cli/): The practical Unix shell — navigating the filesystem, working with files, grep and find, pipes and redirection, text processing (sort/uniq/cut/awk), permissions, processes and jobs, environment and PATH, ssh/scp/curl, and shell scripting. Data & Stats: - [NumPy](https://datarekha.com/numpy/): The array engine under Pandas, scikit-learn, PyTorch, and JAX. Vectorization, broadcasting, linear algebra. - [Pandas](https://datarekha.com/pandas/): The de-facto Python tool for tabular data. Reading, cleaning, reshaping, joining, GroupBy, time series. - [Business Analytics](https://datarekha.com/business-analytics/): The analytics an MBA pays for — unit economics (CAC, LTV), reading a P&L, customer segmentation (RFM), funnels and cohorts, decision analysis (expected value, decision trees), forecasting, optimization, A/B testing, and storytelling with data. Every business term defined from scratch, beginner to advanced. - [Math for ML](https://datarekha.com/math-for-ml/): The math an ML engineer actually uses. Linear algebra, calculus, probability and statistics — including Simpson's paradox and KL divergence — explained with NumPy code. EDA & Visualization: - [Storytelling with Visualisation](https://datarekha.com/visualization/): Matplotlib and Seaborn for exploratory analysis and ML diagnostics — plus the craft of data storytelling: choosing the right chart, the narrative arc of a chart, annotation and guiding the eye, and colour and accessibility for clarity. ML Fundamentals: - [Machine Learning](https://datarekha.com/ml/): Classical ML with scikit-learn, XGBoost, LightGBM. Regression, classification, trees, evaluation, data leakage, hyperparameter tuning, model calibration, and the curse of dimensionality. Advanced ML: - [Time Series](https://datarekha.com/time-series/): Forecasting done right — trend/seasonality decomposition, stationarity and differencing, ACF/PACF, AR and MA, ARIMA, SARIMA, SARIMAX, VAR, exponential smoothing and Holt-Winters, Prophet, lag and rolling features, and leakage-free walk-forward validation. - [Recommender Systems](https://datarekha.com/recsys/): How Netflix, Spotify, and Amazon decide what's next — the utility matrix, content-based filtering, user- and item-based collaborative filtering, similarity metrics, matrix factorization (SVD/ALS), implicit vs explicit feedback, ranking metrics (precision@k, NDCG), the cold-start problem, and hybrid and neural (two-tower) recommenders. Deep Learning: - [Deep Learning](https://datarekha.com/deep-learning/): PyTorch-first. Autograd, optimizers, loss functions, regularization, transformers (self-attention, multi-head, positional encodings), Hugging Face, LoRA/QLoRA fine-tuning, and generative models (GANs, VAEs, diffusion). MLOps & Production: - [MLOps](https://datarekha.com/mlops/): Ship models, not notebooks. Lifecycle, MLflow, Docker, CI/CD, FastAPI/BentoML serving, drift monitoring, incident response. - [PySpark](https://datarekha.com/pyspark/): Distributed compute at scale. Hadoop background, Spark internals (driver/executors, DAG, Catalyst, AQE), DataFrame API, production tuning. AI: - [Generative AI](https://datarekha.com/gen-ai/): LLMs in practice. Autoregressive generation, sampling, structured outputs, prompting, tool use, RAG (basic to advanced), evals, cost/latency engineering, self-hosting, quantization, mixture-of-experts, and speculative decoding. - [Agentic AI](https://datarekha.com/agentic-ai/): Production agents. Design patterns (reflection, tool use, planning, multi-agent), the agent-interoperability protocol stack — MCP (agent-to-tools) plus A2A, ACP, and ANP (agent-to-agent) — LangChain, LangGraph, Microsoft Agent Framework, Google ADK, and production frameworks: Mem0 (agent memory), FastMCP (build MCP servers fast), and Firecrawl (web data for LLMs and RAG). Exam Prep: - [GATE DA](https://datarekha.com/gate-da/): Exam-grade Data Science and Artificial Intelligence for GATE DA: probability, statistics, linear algebra, calculus, Python, DSA, DBMS, warehousing, machine learning, and AI search/reasoning. ## Learning paths - [Data Engineer](https://datarekha.com/paths/data-engineer/): Python → SQL → Pandas → PySpark → MLOps basics. - [ML Engineer](https://datarekha.com/paths/ml-engineer/): Python → NumPy → Pandas → Math for ML → ML → Deep Learning → MLOps. - [AI / LLM App Builder](https://datarekha.com/paths/ai-app-builder/): Python (async, Pydantic, FastAPI) → SQL essentials → Generative AI → Agentic AI. - [Data Scientist](https://datarekha.com/paths/data-scientist/): Python → Pandas → Visualization → SQL → Math for ML → ML. - [Business Analyst](https://datarekha.com/paths/business-analyst/): Business Analytics → SQL essentials → Visualization → Stats (probability & A/B testing). - [MLOps / Platform](https://datarekha.com/paths/mlops-platform/): Python → MLOps → PySpark. - [Research-leaning](https://datarekha.com/paths/research-leaning/): Math for ML → Deep Learning → Generative AI. ## Blog - [Blog](https://datarekha.com/blog/): Long-form pieces on production AI, GATE DA, and career growth — how Anthropic, Cursor, Devin, Perplexity, and Notion actually deploy agents, deep GATE Data Science and Artificial Intelligence essays, 50 corporate soft-skills field notes (communication, teamwork, productivity, career growth), plus a data-backed series on the modern workplace and employee wellbeing — the global engagement crisis, burnout and quiet cracking, AI's effect on jobs and after-hours work (the "AI velocity paradox"), financial stress, workplace loneliness, the four-day week, and whether corporate wellbeing spending actually works — written for data and engineering professionals and grounded in Gallup, WHO/ILO, Oxford, PwC, Mercer, Deloitte, and Nature studies. Categories: Agents, RAG, Frameworks (LangGraph, MAF, ADK, MCP), Infrastructure, Cloud, Patterns, GATE DA, Career. ## Tools - [Interview Q&A](https://datarekha.com/interview/): 500+ data science, SQL, Python, statistics, machine learning, deep learning, NLP/LLM, MLOps, data engineering, and case/behavioral interview questions with worked expert answers, diagrams, the common trap, and the role each is asked for (Data Analyst, Data Scientist, ML Engineer, Data Engineer, AI Engineer, MLOps). Each question is its own page with QAPage structured data; topic hubs at /interview/sql/, /interview/machine-learning/, etc. - [Glossary](https://datarekha.com/glossary/): A plain-English glossary of ~190 data science, statistics, machine learning, deep learning, and AI terms (overfitting, p-value, transformer, RAG, window function, …), each linked to the lesson that teaches it, with DefinedTermSet structured data. - [Cheat sheets](https://datarekha.com/cheatsheets/): Dense one-page references for Python, SQL, Pandas, NumPy, Git, and scikit-learn — the syntax and commands used daily. - [References](https://datarekha.com/references/): The reference hub — glossary, cheat sheets, interview Q&A, and quick-access section indexes. - [FAQ](https://datarekha.com/faq/): Straight, accurate answers to the questions people most commonly ask across every subject — Python, SQL, Pandas, PySpark, machine learning, deep learning, MLOps, generative AI, agentic AI, DSA, business analytics, and GATE DA. Also surfaced as per-section FAQ blocks with FAQPage structured data. - [Search](https://datarekha.com/search/): Client-side search across every lesson. ## Full index - [Full content map](https://datarekha.com/llms-full.txt): every published lesson and blog post, grouped by subject and chapter, with links. - [Sitemap](https://datarekha.com/sitemap-index.xml) ## About - [Privacy policy](https://datarekha.com/privacy/): What we collect (essentially: email if you sign in, completed lesson refs). - [Terms of service](https://datarekha.com/terms/) - [About](https://datarekha.com/about/)