I have had to learn things properly as an adult more than once: a new subfield when I moved from my doctoral work into industry, a new stack when I decided to build an Android app on my own, a new regulatory vocabulary when the products I work on became things that get reviewed by people who do not care about model architecture. None of it went the way the course catalogues imply.
This is an article about how technical skill is actually acquired by a working professional with limited evenings, and about why so much of the "upskilling revolution" content is selling a product rather than describing a process.
The market and its incentives
The alternative-credential industry is large and growing, and its growth is genuinely connected to a real shift: surveys of employers show large numbers dropping formal degree requirements for some roles, and there is real appetite for people who can demonstrate capability directly.
But notice what the industry sells and how it is measured. Course platforms are measured on enrolments and completions. Bootcamps are measured on placement rates that they compute themselves, usually with definitions that exclude inconvenient categories. The whole apparatus is optimised to sell you a start, because starting is the transaction. Nothing in the business model depends on you being good at the thing a year later.
Two things follow. First, treat any specific salary figure attached to a specific credential as marketing. The effect of a certificate on earnings depends on the role, the market and everything else you bring, and nobody has clean causal data on it. Second, be sceptical of the bootcamp economics specifically. The sector's outcomes have deteriorated sharply from its 2017-2021 peak, several prominent schools have collapsed or been penalised by regulators over inflated placement claims, and income-share agreements — sold as risk-free — have repeatedly turned out to carry more risk to the student than advertised. Some programmes are fine. The category is not a safe default, and the era when a three-month course reliably ended in a well-paid job is over.
What actually builds skill
The honest version is unglamorous and hard to sell, which is why you rarely read it.
You need a real problem, not a syllabus. The single strongest predictor of whether I have retained something is whether I was trying to accomplish something specific with it. Courses teach in the order the material is logically organised; problems teach in the order that things break, which is the order that sticks. When I was building my budgeting app, I learned more about Android's lifecycle from one afternoon of a bug that only appeared on rotation than from any amount of reading, because the reading had nothing to attach to.
You need to produce something someone else can see. Output forces closure. It is the difference between having watched a lecture on a topic and having written the paragraph explaining it, where you discover that the thing you thought you understood has a hole in the middle. A deployed project, a written explainer, a merged pull request, an internal talk — the medium matters less than the fact that it leaves your head.
You need feedback from someone who knows more than you. This is the part that is genuinely hard to get outside a job or a good community, and it is the main legitimate advantage of a structured programme. Code review, a mentor, an active project's maintainers, even a strong community forum. Without it, you plateau at a level defined by your own blind spots and do not notice.
You need to do it repeatedly over a long time. Skill acquisition is not a sprint you complete; it is a habit that accretes. Five hours a week for a year beats forty hours a week for six weeks, by a large margin, because the intervals are where consolidation happens — and because the sprint version ends and the habit version does not.
And you need to allow it to be unpleasant. The productive state is confusion. Courses are engineered to minimise it, which is why they feel good and transfer poorly. If your learning feels smooth, you are probably revising something you already know.
Course-collecting, and how to notice you are doing it
A specific failure mode worth naming: accumulating completed courses as a substitute for doing anything. It feels like progress, it produces certificates, and it can go on for years.
The diagnostic is simple. In the last three months, what have you built, written or shipped using the thing you are learning? If the answer is nothing, more courses will not help, and the problem is not that you have not found the right one.
The related error is over-planning the curriculum. People spend weeks constructing the perfect learning path — comparing platforms, sequencing prerequisites, building a spreadsheet. This is procrastination with good posture. Pick an adequate resource and a small real project, and correct course as you go.
Choosing what to learn
The demand rankings that circulate are mostly noise, and following them is a bad strategy anyway, because by the time a skill is on a widely-circulated list, the easy positions are taken and you are competing with everyone else who read the list.
Better filters, in the order I would apply them.
What does your current work almost require? The highest-return learning is usually adjacent to what you already do, because you have context, you have a real problem to attach it to, and you can apply it immediately at work where someone will review it. This is boring advice and it is right.
What compounds? Some skills decay fast and some are still paying out decades later. Specific framework versions decay. Statistics, writing clearly, understanding how systems fail, the fundamentals of whatever domain you work in — these compound. Weight your time accordingly, while accepting that you need some of the perishable kind to stay employable.
What do you actually find interesting enough to still be doing in six months? People underrate this. The sustainability of the habit dominates everything else, and you cannot force sustained voluntary effort on a topic that bores you.
Use your employer's budget, properly
Many companies offer a learning stipend, and most of it goes unclaimed or gets spent on a course nobody finishes. Ask what is available; the answer is frequently more than the official policy suggests, particularly for conference attendance, books and time.
The most valuable thing to negotiate for is not money but scope. Getting assigned to a project that requires the skill you want is worth more than any course, because it comes with time, review and consequences. Asking a manager for that is a normal conversation and most of us are glad to have it — a person who wants to grow in a direction the team needs is the easiest thing in the world to say yes to. Ask for the project, not the voucher.
The honest expectation
Learning a genuinely new technical skill to an employable standard, part-time, alongside a job, takes most people somewhere between six months and two years. It is punctuated by long stretches where you are not visibly improving. Nobody structures it for you, and the credential at the end matters considerably less than the four things you built along the way.
That is a worse offer than the one in the advertisements. It also happens to work, which the advertised version frequently does not.
Tags
Taresh Sharan
support@sharaninitiatives.com