How I build, start to finish.
Deep dives into products and automations I designed, built and run: the problem, the decisions, the architecture and the results.
A SQL platform where every query runs on a real database.
Most SQL tutorials stop at reading. I built a platform where learners write real queries against live PostgreSQL sandboxes and get graded in milliseconds.
The problem
People could follow along with tutorials, but JOINs, CTEs and window functions stayed fuzzy until they wrote queries themselves and saw immediate, honest feedback. Learners asked for two things above all:
“I wanted clear tasks I could complete in sequence, then repeat with confidence. Joins were still the hardest part.”
Thea, LearnWithSQL learner
“I needed logical reasoning and plain-language explanations for each command so I could remember it long term.”
Sarita, LearnWithSQL learner
The approach
Short lesson, then write the query, then see it graded. Every exercise has an explicit goal and expected output shape, explanations teach the reasoning behind each clause, and learners can re-run, take hints and repeat until it sticks.
| first_name | salary |
|---|---|
| Michael | 13,000 |
| Shelley | 12,000 |
| Alexander | 9,000 |
| Course | Level | Lessons | Exercises | Minutes |
|---|---|---|---|---|
| SQL Fundamentals | Beginner | 5 | 19 | ~63 |
| Filtering & Logic | Beginner | 4 | 14 | ~48 |
| Aggregations & Grouping | Intermediate | 4 | 14 | ~48 |
| Joins | Intermediate | 4 | 14 | ~48 |
| Subqueries & CTEs | Intermediate | 4 | 13 | ~46 |
| Window Functions | Advanced | 4 | 12 | ~44 |
Real Python in the browser, with zero installs.
Reading about code feels like progress and rarely is. LearnWithPython turns the opposite idea into a system: meet a concept, run it immediately, build something with it, then review it before it fades.
› Running hello.py
Hello, world! (1)
Hello, world! (2)
Hello, world! (3)
The problem
Beginners get stuck on setup before writing a line, and video courses build familiarity instead of skill. The fix had to remove installs completely and make every lesson end in code that runs.
The approach
A full CPython interpreter runs locally in the browser through WebAssembly (Pyodide), so what works there works on your machine. Lessons are short and each ends with a live challenge graded instantly. A Smart Review engine turns concepts into flashcards and reschedules them with spaced repetition, and XP, levels, streaks and achievements make progress visible.
The path
Ten connected courses go from print("Hello") through data structures, OOP, algorithms, files and REST APIs to expert Python: generators, decorators, context managers and the data model. The APIs course uses the site’s own backend as its case study.
An intake form that answers itself.
A working pattern I use for client automations: a web form triggers an AI workflow that reasons about the request and emails back a complete answer, with no human in the loop.
Structured form
Vehicle year, make, model, mileage, what happened and the desired outcome.
Make webhook
JavaScript posts the answers as JSON to a Make scenario the moment it’s submitted.
AI prompt chain
Prompts primed with mechanic knowledge work through likely causes and next steps.
Email answer
A written, personalized reply lands in the user’s inbox.
Where this pattern pays off for businesses
Quote requests, support triage, lead qualification and intake for service businesses. The same architecture powers the lead-to-booking systems ScalingThis builds for clients.