Skip to content

Analytics Made Simple

Learn

Step-by-step guides for SQL, tools, and practical AI — skills you can use at work. Learn by doing, not by jargon.

Where do you want to start today?
Showing all 51 series across 5 tracks

🎓 Core Analytics & Data Skills

The foundational curriculum: go from messy spreadsheets to confident SQL, Python, charts, and metrics you reuse everywhere.

Track 01 · 7 Series
★ Flagship Core Curriculum

SQL series

Step-by-step SQL — from playground setup through joins, subqueries, and optimization.

Key Milestones:
  • Parts 1–4: Playground setup, SELECT, filtering, and sorting
  • Parts 5–8: Grouping, aggregates, multi-table joins & subqueries
  • Parts 9–14: Views, indexes, window functions, CTEs & query tuning
Start Part 1 (14 Lessons) →
SELECT department, COUNT(order_id) AS volume, ROUND(AVG(revenue), 2) AS ticket FROM warehouse.orders WHERE status = 'shipped' GROUP BY 1 HAVING volume > 1000;
 Visualization Intermediate Charts that make sense Make charts that answer decisions: purpose, chart choice, honest axes, accessible color, dashboards vs slides, annotation, and chart crime fixes. 7 parts Start series → Quality Intermediate Data quality for people who ship numbers Hands-on data quality for analysts who ship numbers: dimensions, profiling, dedupe, standards, time, checks, and a light scorecard. 7 parts Start series → Spreadsheets Beginner From spreadsheets to real data Excel and Sheets habits that scale: when a file becomes a liability, real tables, keys, cleaning, clean exports, and a one-page table contract. 6 parts Start series → Metrics Intermediate Metrics that matter From goals to metric specs, leading signals, North Star design, incentive traps, and review meetings that actually decide something. 6 parts Start series → Foundations Beginner Analytics foundations Questions before data — learn to define the problem so every chart and query actually helps someone decide. 5 parts Start series → Python Intermediate Python for analytics Clean data, explore with pandas, and turn notebooks into reusable analysis you can trust at work. 12 parts Start series →

🤖 Everyday AI for Work & Life

Practical AI workflows for knowledge workers — prompting, memory, personal assistants, and real work tasks.

Track 02 · 8 Series
★ Flagship Workplace AI

Practical AI for analytics people

A nine-part series for analysts: what AI can and cannot do, tokens, prompts, evals, RAG, agents, privacy, documentation, and a personal ops checklist.

Key Milestones:
  • Parts 1–3: Capabilities, limits, token economics & prompt framing
  • Parts 4–6: RAG foundations, agents, evals & structured data extraction
  • Parts 7–9: Privacy boundaries, governance checklists & production ops
Start Part 1 (9 Lessons) →
// Workplace AI Loop Input: messy business brief Step 1: Extract constraints & entities Step 2: Compare model reasoning traces Step 3: Verify outputs with facts Result: Decisions shipped 5x faster
 Agents Intermediate AI agents for everyone An agent is a chat that can take steps. Watch, approve, or walk away on purpose. Check the work without being an engineer. 5 parts Start series → Safety Beginner AI safety for everyday life Scams, deepfakes, school honesty, health and money limits, kids on shared devices, and the job skills that still matter. 5 parts Start series → AI Setup Beginner AI setup from zero Accounts, payments, apps, and where your files go. Start on Free. Pay on the website. One vendor at a time. 5 parts Start series → Assistants Intermediate Everyday AI assistants beyond chat boxes Explore the shift from reactive chat boxes to proactive assistants and autonomous personal agents: mental models, permissions, OpenClaw, Hermes Agent, Grokbot, and real-world workflows. 8 parts Start series → Workflows Beginner Memory, projects, and files across AI tools Chat history is not memory. Projects are not a folder. Learn where uploads go and how to turn a messy directory into a working project. 4 parts Start series → Prompting Beginner Prompting for everyone Talk to AI like a clear human. Context, constraints, examples, iteration, and when to stop prompting and do the work yourself. 5 parts Start series → Tool Chooser Beginner Which AI product should I use? Pick a chat box, a coding agent, an office helper, or a local model by the job, not the brand name. 6 parts Start series →

🛠️ Major AI Platform Masterclasses

Deep-dive operational playbooks for frontier models and developer ecosystems with official marks.

Track 03 · 10 Series
★ Frontier Model Masterclass

Claude

Anthropic Claude from first login through product map, Claude Code, and Cowork — one continuous path.

Key Milestones:
  • Parts 1–3: Prompt caching, Projects, Artifacts & context hygiene
  • Parts 4–6: Claude Code CLI tool execution, subagents & workflows
  • Parts 7–10: MCP server integration, tool boundaries & developer best practices
Start Part 1 (31 Lessons) →
$ claude > Analyze Q3 churn data and propose 3 SQL fixes Thinking Process (Extended Thinking)… [Tool Call] read_file query.sql [Result] 3 queries optimized Ready for review.
 OpenAI Intermediate ChatGPT OpenAI ChatGPT from everyday chat through product map, Work mode, Codex, and Custom GPTs — one continuous path. 30 parts Start series → Google Intermediate Gemini Google Gemini from first login through product map, Workspace, and coding surfaces — one continuous path. 26 parts Start series → xAI Intermediate Grok A 24-part path into xAI Grok: getting started, product map (chat vs Build vs Imagine vs API), everyday writing and verification, Grok Build in the terminal, and Imagine for images and short video. 24 parts Start series → Meta Intermediate Meta Llama from scratch Llama is a family of open-weight models from Meta, not a chatbot app. Map the family, the license, and when a $20 closed chat is easier. 7 parts Start series → DeepSeek Intermediate DeepSeek from scratch DeepSeek is a Hangzhou lab with a free chat app and open weights. Separate the product from the files you can download. 7 parts Start series → Alibaba Intermediate Qwen from scratch Qwen is Alibaba's model family: chat surfaces, open weights, and a huge multilingual line. Start with the map, not the zoo. 6 parts Start series → Moonshot Intermediate Kimi from scratch Kimi is Moonshot's assistant, famous for long context. The app, the K2/K3 weights, and what long context actually buys you. 6 parts Start series → Z.ai Intermediate GLM from scratch GLM is Zhipu's (Z.ai) family, including the GLM-5.x era. ChatGLM heritage, chat.z.ai, open weights, coding plans. 7 parts Start series → Hugging Face Intermediate Learn Hugging Face from scratch A comprehensive, zero-fluff guide to navigating and mastering the Hugging Face ecosystem: understanding open weights and GGUFs, evaluating model cards and licenses, inspecting datasets, launching interactive Spaces without code, deploying local and serverless inference, and establishing bulletproof model safety practices. 8 parts Start series →

🧠 Open-Source & AI Engineering

Run, build, fine-tune, and inspect models locally — llama.cpp, Unsloth, RAG, vector databases, and autonomous harnesses.

Track 04 · 14 Series
★ Deep AI Engineering Path

Fine-tuning open models with Unsloth from scratch

A comprehensive, zero-fluff engineering guide to fine-tuning open foundation models using Unsloth. Understand the mechanics of weight updates versus prompt engineering and RAG, master manual CUDA autograd kernels that deliver 5x training speedups with 80% lower VRAM, explore LoRA and QLoRA low-rank adapter mathematics, prepare high-quality Alpaca and ShareGPT instruction datasets, fine-tune Llama 3 and Qwen models on free Google Colab GPUs, and export merged adapters directly to GGUF, Ollama, and vLLM for high-throughput production serving.

Key Milestones:
  • Part 1: Weight updates vs RAG & prompt engineering trade-offs
  • Part 2: 5x faster custom CUDA autograd kernels with 80% less VRAM
  • Part 3: LoRA / QLoRA training on Colab GPUs & GGUF production export
Start Part 1 (3 Lessons) →
from unsloth import FastLanguageModel model, tokenizer = FastLanguageModel.from_pretrained( model_name = 'unsloth/Meta-Llama-3.1-8B', max_seq_length = 2048, load_in_4bit = True, ) # 5x faster, 80% less memory
 Meta Advanced llama.cpp from scratch Master Georgi Gerganov's bare-metal C++ inference engine: from zero-dependency builds and GGUF model execution to local OpenAI-compatible API servers and production optimization. 7 parts Start series → Agent Harnesses Advanced AI harnesses and coding agents from scratch Understand the runtime loops, tool execution boundaries, context packing algorithms, and evaluation frameworks behind modern coding agents and autonomous AI harnesses. 8 parts Start series → Automation Intermediate Automation platforms from scratch (n8n, Zapier, and friends) Connect your apps without writing throwaway code: learn triggers, webhooks, visual pipelines, n8n self-hosting, Zapier patterns, AI steps, and production reliability. 10 parts Start series → LlamaIndex Advanced LlamaIndex from scratch Learn LlamaIndex from the ground up: discover how data connectors, document nodes, vector indexes, query engines, and agentic tools turn messy files into production-grade knowledge pipelines. 7 parts Start series → Local AI Intermediate Local LLMs from scratch Run powerful artificial intelligence models directly on your machine with zero cloud telemetry, complete data privacy, and zero monthly subscription costs. 9 parts Start series → Open Source Intermediate Open-source AI explained What open means for models: weights, licenses, and the homework of running things yourself. Not a free lunch. 10 parts Start series → Model Maps Intermediate OSS and open-weight model map The open-weight model ecosystem decoded: foundations, fine-tunes, -Instruct variants, small edge runtimes, and practical evaluation benchmarks. 6 parts Start series → RAG Advanced RAG from scratch Master Retrieval-Augmented Generation (RAG) from the ground up: learn how document chunking, vector embeddings, metadata filters, hybrid search, and citation runtimes ground language models in live, verifiable facts. 8 parts Start series → Open Models Intermediate Run open models from scratch Hosted open-model chat, desktop runners, one local stack, offline tasks, hardware, updates, and sharing a machine with family. 7 parts Start series → Vector DBs Advanced Vector database products map Demystify the vector database ecosystem: explore in-memory libraries vs dedicated engines, pgvector tradeoffs, managed vector platforms, hybrid search algorithms, and exact memory sizing math. 6 parts Start series → Video AI Advanced Video models from scratch From text-to-video and image-to-video diffusion architectures to cinematographic prompting briefs, credit cost optimization, deepfake governance, and production hybrid workflows. Master generative video models without burning budget or falling for synthetic hype. 6 parts Start series → Vision & OCR Advanced Vision, OCR, and document extraction from scratch A comprehensive practical guide to optical character recognition, document AI pipelines, vision language models, and extracting structured data from paper, scans, and PDFs. 8 parts Start series → Voice AI Advanced Voice models from scratch From text-to-speech (TTS) neural vocoders and speech-to-text (STT) Whisper architectures to zero-shot voice cloning, latency optimization, conversational agents, ethics, consent, and hybrid audio engineering. Master practical generative voice without synthetic uncanny valley fails. 1 part Start series →

🏢 Data Systems, Engineering & Applied Practice

How data moves, scales, and delivers value: pipelines, dbt, Airflow, stewardship, BI, and domain analytics.

Track 05 · 12 Series
★ Production Systems Anchor

How data actually moves

A seven-part series for analysts who inherit pipelines: path, latency, storage homes, orchestration, dbt concepts, environments, and observability.

Key Milestones:
  • Parts 1–2: Trace data movement, lineage & latency requirements
  • Parts 3–5: Storage tiers, orchestration engines & dbt transformations
  • Parts 6–7: Multi-environment deployments & proactive observability alerts
Start Part 1 (7 Lessons) →
Source: CRM & Event Streams ↓ (Extract & Load: 15m cadence) Raw Warehouse Landing ↓ (dbt Transformations & Tests) Curated Dimensional Marts ↓ (Power BI / Metric APIs) Executive Dashboards (99.9% SLA)
 Apache Advanced Airflow hello DAG First DAG, retries, sensors, and when not to use Airflow. 2 parts Start series → Career Beginner Analyst career path IC ladder, portfolio, and storytelling toward senior trust. 3 parts Start series → Customers Beginner Customer analytics basics Funnels, conversion math, and cohorts for product analysts. 3 parts Start series → Governance Beginner Data stewardship at work A six-part series on weekly data habits: roles, catalogs, access, incidents, retention, and working with Legal and Security without panic. 6 parts Start series → dbt Labs Advanced dbt project lab Hands-on layout, models, tests, and PR review. 3 parts Start series → Experiments Intermediate Experimentation culture Design, guardrails, and decision reviews beyond pure stats. 3 parts Start series → Finance Intermediate Finance analytics for non-finance ARR, churn, and cohort retention with careful definitions. 3 parts Start series → Geospatial Beginner Geospatial for beginners When maps help and how to avoid misleading them. 2 parts Start series → Data Culture Beginner Inclusive data products Accessibility and whose story your metrics miss. 2 parts Start series → Microsoft Intermediate Power BI starter Model, measures, and trustworthy publish for beginners. 3 parts Start series → Statistics Intermediate Statistics for analysts Bias, sampling, uncertainty, and A/B intuition without the math degree. 4 parts Start series →
  • ,

    Claude Cowork: what it is for

    Claude Cowork is the agent surface for non-coding knowledge work: folders and tools you choose, multi-step tasks, deliverables you still review. Part 3 of the Claude product map…

    Claude Cowork: what it is for
  • ,

    Custom GPTs vs plain chat

    Part 2 of ChatGPT product map. When a Custom GPT earns its keep versus plain chat: saved instructions, knowledge files, plan limits on creating GPTs, and why a…

    Custom GPTs vs plain chat
  • ,

    DataFrames as tables

    A pandas DataFrame is just a table with rows, columns, and types you can inspect. Learn to load a CSV, check shape and dtypes, rename columns, and map…

    DataFrames as tables
  • ,

    Claude Code: what it is for

    Part 2 of Claude product map. What Claude Code is for when you will touch a project: explore a repo, edit many files, run tests, and open PRs,…

    Claude Code: what it is for
  • ,

    Chat vs Work vs Codex

    Part 1 of ChatGPT product map. Plain English on Chat, Work, and Codex: what each mode is for, how usage differs, and a chooser you can use before…

    Chat vs Work vs Codex
  • ,

    Setup without tears

    Set up Python for analytics without the folklore that scares people off. Create a virtual environment, install pandas with pip, and load a tiny CSV so your first…

    Setup without tears
  • ,

    Claude.ai chat vs everything else

    Part 1 of Claude product map. A plain-English tour of Claude chat versus Code, Cowork, Design, Science, and the API Console so you know which door to open…

    Claude.ai chat vs everything else
  • ,

    Privacy, work rules, and when not to use it

    Privacy, company policy, and the right tool matter more than clever prompts. Use this series-end guide for when not to use ChatGPT, a good-enough checklist, and what to…

    Privacy, work rules, and when not to use it
  • ,

    Why Python for analytics (and when to stay in SQL or Sheets)

    Python is optional for many analytics jobs, and that is a feature, not a failure. Learn when Sheets or SQL still win, when a notebook saves real hours,…

    Why Python for analytics (and when to stay in SQL or Sheets)
  • ,

    When Claude is the wrong tool

    Closes Learn Claude from scratch with judgment: four wrong-tool zones, a good-enough checklist, walk-away signals, and what comes next in the product map, Code, and Cowork series.

    When Claude is the wrong tool
  • ,

    When Grok is the wrong tool

    Generative AI tools like Grok are powerful assistants, but they can be dangerous when misapplied. In this final part of the series, we cover the six critical danger…

    When Grok is the wrong tool
  • ,

    Connectors and apps (Drive, calendars, etc.)

    Connectors and apps link ChatGPT to Drive, mail, Slack, CRMs, and more. Climb from plain chat to uploads, read-only access, write actions, and scheduled work only as far…

    Connectors and apps (Drive, calendars, etc.)