Start here
Start with one public lesson. Build one proof trail.
If AI engineering feels too large, make the first decision small: read lesson 01, touch four real tools, keep evidence of what changed, then decide if support is useful.
- First action
- Lesson 01
- Proof
- One trail
- Support
- Later
First tool path
Your first day is not abstract. You touch real engineering surfaces.
Before the curriculum becomes a full-stack product, it starts with four simple surfaces: ask AI for help, read code in an editor, run the lesson in the browser, and save proof of work in GitHub. These are the same surfaces that appear again throughout all 20 lessons.
Ask
AI pair
Use AI assistance as a draft partner, then learn the habit of checking what it produced.
Read
Editor workspace
Open code in a real editor, move through files, and understand the project as a workspace.
Run
Browser code
See JavaScript turn into visible behavior so code stops feeling like hidden text.
Prove
GitHub repository
Publish work where it can be inspected, reviewed, and improved over time.
Why these first tools matter
The beginner path becomes professional vocabulary later.
On day one, these tools feel like names on a screen. By lesson twenty, they become engineering vocabulary you can explain: editor, repository, runtime, deployment, API, model provider, structured data, and documentation. That vocabulary is what makes your work easier for another engineer to inspect.
Platform and company logos are ecosystem references only. They do not imply partnership, endorsement, interview access, hiring promise, salary promise, or placement guarantee. A beginner earns credibility by building inspectable work, not by seeing a logo on a page.
Day 01
AI pair plus editor
You begin by asking for help, then reading the generated code in a real workspace instead of copying blindly.
- AI pair
- editor
- code review
Proof
Repository habit
The early GitHub habit grows into visible proof: commits, README notes, screenshots, limitations, and a project trail.
- GitHub
- Git
- README
Ship
Runtime to public demo
Later lessons turn local code into a running system: Node.js runtime, Docker packaging, and Cloudflare reachability checks.
- Node.js
- Docker
- Cloudflare
AI app
Model boundary thinking
AI becomes product work when requests, responses, types, and costs are handled as ordinary engineering boundaries.
- OpenAI
- JSON
- TypeScript
Common fears, named
The reasons you think this is not for you, addressed.
Most of the people who would benefit from this curriculum talk themselves out of starting. The reasons are predictable. Here is the honest answer to each.
"I am not from a science background"
You do not need PCM, calculus, or a CS degree.
The curriculum starts at "what is code" in lesson 01. No mathematical prerequisites. No physics, chemistry, or biology assumed. Commerce, arts, and humanities students can inspect the public syllabus before deciding whether the engineering path fits them.
"I have never coded before"
Lesson 01 explicitly assumes zero prior coding.
You will use an AI coding assistant to scaffold real code while learning what the tool is doing. By lesson 03 you will have built a small browser game. The "never coded" state is the starting state.
"I have a full-time job / college and no time"
The curriculum is calibrated for 5–8 hours per week.
One lesson per week is a reasonable pace. Twenty lessons = roughly twenty weeks at one per week, or three to four months at two per week. Most learners study evenings and weekends. No 12-hour days required.
"I am too old to start"
There is no age cap on engineering.
The curriculum itself has no age gate. Working professionals, career switchers, and parents returning to work can study at the same public starting point as anyone else. The honest filter is sustained practice, not birth year.
"My English is not great"
You can read the lessons in your own time.
The lessons are written in straightforward English. Read at your own pace. Engineering vocabulary is universal — the lessons explain every term before using it, and the glossary is always there when a word blocks you.
"I cannot pay for support yet"
Start with the public syllabus before you decide anything paid.
Twenty lessons, 33 articles, the glossary, and the reading paths form a public proof layer. It is complete enough to judge the teaching quality before you ever contact anyone.
Your first 30 minutes
A concrete first half-hour with this curriculum.
The hardest moment in learning anything is the very first action. To make that easier, here is a specific, time-boxed first 30 minutes you can do right now without installing anything, without signing up, and without any prior knowledge.
- 0:00 – 0:05Read
Read the homepage and the why
Open the homepage and the /why-abcsteps page in two browser tabs. Read both. Five minutes. The goal is to know what this is, what it is not, and whether it fits you.
- 0:05 – 0:15Open
Open lesson 01 — AI-Assisted Code
Click into the first lesson. Read the "Before You Study" section and "The Concept" section. Do not try to do anything yet. Just read what the lesson is about. Ten minutes.
- 0:15 – 0:25See
Open the AI/editor tool context
Open the AI assistant or editor context named in the lesson. If you do not have that exact tool yet, use the screenshots and notes to understand what the surface is for. Type one small question and watch how the answer is structured.
- 0:25 – 0:30Decide
Decide if you want to continue
Five minutes of honest self-reflection. Was that interesting? Could you see yourself doing more of it? If yes, continue with the rest of lesson 01 and the lab. If no, that is also useful information — engineering is not for everyone, and finding out cheaply is better than finding out after spending lakhs on a degree.
That is the first 30 minutes. If you got through them, you have a concrete starting signal: one page read, one tool surface inspected, and one decision about whether to continue.
The rhythm
How a serious learner studies this curriculum.
The curriculum is calibrated for a sustainable pace. Not "complete in 30 days" hype, not 12-hour days that burn you out — a working rhythm a person with a job or college schedule can keep for months.
- 01
One lesson per week
Aim for one lesson per week as the steady pace. Some weeks you will do more; some less. The goal is consistency over speed.
- 02
Read first, build second
Read the entire prepare.md before you touch any tool. The reading is fast; the building is slow. Doing them in the right order saves hours.
- 03
Build the lab even badly
Each lesson has a lab. Do it even if your version is uglier than the example. A bad version of the project teaches more than reading three more lessons without building.
- 04
Review at the end
Read the review.md. Try the quick quiz. Skim the bonus challenge. The review locks the lesson into long-term memory and points you at next steps.
- 05
Repeat for 20 weeks
Twenty lessons at this pace = roughly five months. Five months is a real chunk of time, and the result is a portfolio of working projects you can show.
When you get stuck
Getting stuck is not failure — it is the work.
Every working engineer gets stuck. The skill is not avoiding stuck-ness; it is having a routine for getting unstuck. Here is the routine to use as you go through the curriculum.
Step 1
Re-read the section
Most "I am stuck" moments are "I read too fast" moments. Go back, read the paragraph again with fresh eyes. Often that is the entire fix.
Evidence to keep
Keep the lesson URL and the exact heading where the confusion started.
Step 2
Search the exact error
Copy the error message verbatim and paste it into a search engine or an AI assistant. Engineers do this constantly. It is not cheating; it is the work.
Evidence to keep
Keep the exact error text, the command you ran, and the first answer that helped.
Step 3
Sleep on it
If you have spent more than 30 minutes stuck on the same thing, stop. Come back tomorrow. The brain solves problems while you sleep more reliably than another hour at the keyboard.
Evidence to keep
Write one sentence describing what is failing before you stop for the day.
Step 4
Read the related blog post
Many lessons have a companion blog article linked nearby. The article often explains the concept from a different angle. The angle that clicks is the right one.
Evidence to keep
Save the article route and the paragraph that changed your understanding.
Step 5
Skip and come back
You are allowed to skip a hard step temporarily, finish the rest of the lesson, and come back. Sometimes the rest of the lesson contains the missing piece.
Evidence to keep
Commit or note the last working state before trying the hard step again.
Step 6
Ask a specific question
If the public proof layer still leaves you stuck, send a specific question. Tell Divyanshu which page you are on, what you tried, what happened, and what you expected.
Evidence to keep
Send the route, screenshot, repo or code fragment, command output, and expected behavior.
The public syllabus uses the routine above plus the broader internet: re-read, search the exact error, compare with the related article, and write down what changed. Tool and platform logos are troubleshooting references only: no affiliation, endorsement, interview access, hiring promise, salary promise, or placement guarantee.
Three things to remember
The mindset that gets you to lesson twenty.
Show up, not understand everything.
You will not understand 100% of every lesson on the first read. That is normal. The goal is not perfection on day one — it is showing up consistently. Concepts you do not get in lesson 03 will click in lesson 07.
Build, do not just read.
Every lesson has a lab — a thing to actually build. Reading the lesson without doing the lab is like watching swimming on YouTube. You learn by doing the thing the lesson describes, even badly.
Slow is fine. Stopping is not.
A learner who does one lesson per week for 20 weeks will outpace someone who does ten lessons in a weekend and then quits. Steady beats sprint. The curriculum is calibrated for steady.
When paid guidance helps
Do the first lesson before deciding what to buy.
For a new learner, the best first move is still public: read lesson 01 and keep a proof trail. The paid plan ladder exists for the moment when you know the missing piece is walkthrough, accountability, review, or direct founder attention.
How to choose
Start with the free lessons. Upgrade when videos, Q&A, live accountability, private review, or institutional delivery helps you finish stronger work.
Free start
Free written lessons
INR 0
20 lessons, blog, glossary, and reading paths remain readable without signup, checkout, or account creation.
Open lessonsJob-Ready Track
FoundingABCsteps AI Engineering Course · Job-Ready Track
INR 4,999
A self-motivated learner who has read enough of the public syllabus to trust the teaching style and wants founder guidance without live cohort pressure.
Open Job-Ready TrackCohort
RecommendedCohort — Live group track
INR 14,999
per cohort batch
A learner who stalls in self-paced courses, needs scheduled live accountability, and can attend Saturday/Sunday IST calls.
Open CohortMentorship
Capacity1:1 Mentorship
INR 49,999
per mentee · 8-week engagement
A career switcher, founder, or engineer with stakes high enough that private review is worth more than another course.
Open MentorshipArchitecture Review
Project Architecture Review
INR 24,999
per session
An engineer, founder, or technical lead with a specific decision expensive enough to justify a serious second opinion.
Open Architecture ReviewWorkshop
Institutional Workshop
INR 2L–5L
quoted after scoping
A college, school, bootcamp, or engineering team with 20-60 participants and a clear AI-engineering capability goal.
Open WorkshopReady
Lesson 01 is right here. The first action is one click away.
You have read this page. You know what to expect. You know the rhythm. You know what to do when stuck. The only thing left is the first click.