The self-taught developer's path in India, honestly assessed

Pick one language and stay with it long enough to stop looking up syntax. Python and JavaScript are both defensible starting points; the choice matters far less than the commitment. Switching languages every six weeks is the most common way to spend two years and arrive nowhere.

'Properly' means you can write a two-hundred-line program from a blank file without a tutorial, read an error message and locate the cause, and use the language's standard data structures without thinking. Tutorial completion is not the milestone — building something the tutorial did not cover is.

Expect three to five months at fifteen to twenty hours a week. People who report doing this in four weeks are describing familiarity, not fluency, and the gap shows in the first technical screen.

The transition from learner to developer happens when someone other than you depends on something you built. This can be small: a tool your classmates use for attendance, a script a local business runs weekly, a bot in a community server with fifty members.

Real users generate the experiences that interviews probe for — handling bad input, dealing with someone else's requirement changing mid-build, fixing a production bug under time pressure. No tutorial produces these, and their absence is precisely what makes a portfolio of clone projects read as identical to every other applicant.

Three projects with users beat fifteen without. Deploy them, keep them running, and write a short README that says what problem the thing solves.

• Pick problems where you know the users personally.
• Deploy publicly and keep the link alive.
• Write down one measurable outcome per project.

Indian technical hiring, particularly at product companies and services firms with structured processes, still tests data structures, algorithms, databases and basic systems knowledge. You may disagree with the format; it remains the format. Treating it as an unfair obstacle rather than a subject to study is a decision to be filtered out.

The practical target is roughly two hundred to three hundred well-understood problems across the standard patterns — arrays and strings, hashing, two pointers, trees, graphs, dynamic programming — plus the ability to reason about time and space complexity aloud. Understanding is the operative word; grinding volume without pattern recognition transfers badly.

Alongside that, learn enough SQL to write a join and a group-by without help, enough networking to explain what happens when a URL is entered into a browser, and enough operating systems vocabulary to discuss processes, threads and memory. These come up constantly and are far cheaper to learn than they appear.

The first stall is tutorial dependency — a year of courses, nothing built independently. The exit is to build something unguided and tolerate feeling incompetent for two weeks.

The second stall is the credential wall. Some employers filter on a degree, and no amount of portfolio changes that. The response is not to argue but to route around it: startups, contract work, open-source contributions that lead to referrals, and services companies that hire on tested skill. The filtered employers are a subset, not the market.

The third stall is isolation. Self-taught developers working alone accumulate blind spots they cannot see, because nobody reviews their code. Contributing to an open-source project, however small, is the cheapest available fix — it produces reviewed work, a public history, and people who know your name.

The fourth stall is money. Learning takes twelve to eighteen months at a serious pace, and unpaid learning is a luxury. Freelance work, internships and part-time support or QA roles all buy time and count as experience. Taking one is not a detour.

Course certificates from online platforms carry essentially no weight in technical hiring; they are neither signal nor liability. Cloud certifications (AWS, Azure, GCP associate-level) do carry weight in infrastructure and services hiring, where they map to procurement and partner requirements. Vendor certifications outside cloud are mostly noise.

A public GitHub history with genuine commit activity over a year functions as the strongest available substitute for a degree, because it is difficult to fake and easy to verify. It is more valuable than any certificate you can buy.

Bootcamps vary enormously. The only question worth asking is whether they publish verifiable placement outcomes with company names and salary bands. Those that do not, do not have them.

At fifteen to twenty hours per week, with no prior programming background: three to five months to language fluency, four to six months building real projects in parallel with fundamentals, and three to six months of active applying and interviewing. Twelve to eighteen months total to a first paid role is the honest range, and the first role is frequently below your eventual level.

Full-time study compresses this to eight or nine months, not to three. Anyone quoting three has redefined 'developer job'.

The compounding starts after that first role. Two years of paid work eliminates the degree question at most employers, and the trajectory from there looks the same as everyone else's.