How I Went from SQL Beginner to 119/120 — 40 Practice Exercises That Made the Difference

Deepanshu Garg — Thu, 21 May 2026 16:15:47 +0000

When I started learning SQL, I could write SELECT * FROM table and that was about it.

The problem wasn't understanding concepts — it was getting enough practice to make them stick. Reading tutorials gives you confidence. Writing 40 queries gives you competence.

Last month I scored 119/120 on a SQL capstone. Here's what I learned about practicing SQL effectively:

The Progressive Approach

Don't jump into window functions until you can write a basic JOIN without syntax errors. Here's the progression that worked for me:

Level 1: Basic Queries

Master SELECT, WHERE, ORDER BY, LIMIT. Sounds simple, but 80% of real-world queries are just these with more columns.

Level 2: Aggregation

GROUP BY is where SQL gets interesting. Add HAVING to filter your groups. Use CASE inside aggregations.

Level 3: Joins

INNER JOIN is your daily driver. LEFT JOIN catches everything else. SELF JOIN is weird but powerful once it clicks.

Level 4: Subqueries & CTEs

CTEs (WITH clauses) changed everything for me. Instead of nested subqueries that look like spaghetti, you get clean, readable steps.

Level 5: Window Functions

ROW_NUMBER, RANK, LAG, LEAD, running totals. These are what separate "I know SQL" from "I can write production queries."

Practice > Theory

I compiled 40 exercises following this exact progression into a workbook. Each exercise has:

A self-contained dataset (CREATE TABLE + INSERT — no external files)
Expected output written in comments
A full answer key

If you want it: https://deepanshu647.gumroad.com/l/pkrci

Or if you just want the Level 1 exercises as a free sample, comment below and I'll paste them.

I'm Deepanshu — BTech IT, self-taught data scientist. I document everything I learn.

I Built a 12-Page Data Science Cheat Sheet So I'd Stop Googling Syntax Every 5 Minutes

Deepanshu Garg — Thu, 21 May 2026 13:07:43 +0000

Six months ago, I started a self-directed data science program from scratch.

Month 1 was Excel. Month 2 was SQL. By Month 6, I was fitting Random Forest models and running t-tests in scipy. The problem? I kept jumping between 4 different browser tabs just to remember basic syntax.

So I did what any lazy engineer would do — I put everything into one PDF.

What's Inside

The cheat sheet bundle covers the 4 topics I use daily:

1. Pandas — Data Wrangling

Reading/writing (CSV, Excel, SQL, JSON), filtering, groupby, merge, pivot tables, handling nulls, datetime operations.

2. SQL — Queries, Joins & Aggregations

SELECT structure, WHERE operators, GROUP BY + HAVING, all JOIN types, CTEs, window functions (ROW_NUMBER, RANK, LAG/LEAD), data modification.

3. Scikit-learn — Machine Learning

Train/test split, pipelines, preprocessing (StandardScaler, OneHotEncoder, ColumnTransformer), regression/classification/clustering models, evaluation metrics (accuracy, precision, recall, F1, MAE, RMSE, R2), cross-validation, GridSearchCV, feature importance.

4. Statistics — Foundations & Inference

Descriptive stats, probability distributions, CLT, hypothesis testing, t-tests, ANOVA, chi-square, correlation (Pearson/Spearman), confidence intervals, Type I/II errors, key formulae.

Why a PDF?

I wanted something I could:

Keep open on a second monitor while working
Print and pin to my wall
Search quickly with Ctrl+F
Access offline

Where to Get It

I put the bundle on Gumroad for $12. If you're interested: https://deepanshu647.gumroad.com/l/shcxrw

Or if you just want the Pandas section as a free sample, comment below and I'll send it over.

I'm Deepanshu — BTech IT grad, self-taught data scientist, currently on Day 114 of a 12-month program. Building toward freelance data consulting and an MSc in the Netherlands.

DEV Community: Deepanshu Garg