Programming and Software Development

Our programming pillar makes an argument about how long each kind of knowledge lasts. This hub does something plainer: it draws the territory. Software development covers everything between a stated need and running code, and three readers carve it up in three ways. A practitioner thinks about where the code runs: a browser, a server, a phone, a game engine. A hiring manager thinks in job titles and the stack named in the advert. A syllabus thinks in skills that can be assessed, which is why SFIA lists programming, software design, functional testing, non-functional testing and systems integration as separate skills with their own levels of responsibility, instead of one job called developer.

Twelve topics, seven job families

Four topics are common ground for every developer whatever the contract says: fundamentals, a language, engineering practice and databases. The other eight follow where the software runs or what the team answers for. The table pairs each topic with the role that spends most of its week there and the nearest credential, where one exists. An empty credential cell is information too, because some of the largest parts of the field have no exam anyone asks about.

Full stack is not a thirteenth topic. It is the web and back-end rows held by one person, usually at a company too small to split them. Mobile and game studios hire separately because their toolchains, release channels and performance limits differ enough that a web developer crossing over spends months relearning the ground.

What the credentials prove

Development has fewer respected exams than any other field on this site, and the reason is structural. A language certificate tests recall of syntax, which an interviewer can probe in minutes and a compiler checks all day long; nobody hires on it. The credentials with weight sit where an employer's own systems are at stake. DVA-C02 asks whether your code behaves the way one cloud platform expects, AZ-400 asks how a team moves changes from commit to production, and the ISTQB scheme gives testers a vocabulary shared across employers and countries. The free project-based certificates from freeCodeCamp and Harvard's CS50x prove something narrower but still useful: that you finished substantial graded work. None of them shows that you can join a team, read its codebase and ship a change without breaking anything. A portfolio, a contribution history and the technical interview test that.

Starting points, and what assistants changed

• Career changers with no code behind them begin with fundamentals in one language and stay until they can build something small without a tutorial open. Web development is the usual next step, because the browser gives instant feedback and the job market there is the largest.
• Self-taught developers who already ship tend to have gaps in the same three places: Git beyond the daily commands, automated tests, and how data is modelled in a relational database. Engineering practice, testing and databases are where steady weekly work closes the gap interviewers probe.
• Testers and administrators moving towards development already hold half of a neighbouring topic. Testers know how software breaks, administrators know how it runs, and test automation or CI/CD pipelines are their shortest bridges across.
• Developers aiming at senior roles move into engineering practice and DevOps, where the job shifts from writing features to deciding how the team writes them.

Coding assistants now draft much of the routine code, and that has reshaped the entry level more than the senior one. Boilerplate, simple endpoints and test scaffolding, the tasks juniors used to learn on, are the ones an assistant does fastest. What stays scarce is judgement about a proposed change: whether it is correct, whether it fits the codebase, whether it opened a hole a colleague in application security will find later. The AI-assisted development topic treats that review work as part of the craft.

Neighbouring fields take over at clear edges. Infrastructure as code and the running of cloud platforms live under cloud, with containers and Kubernetes shared at the border. Querying data for analysis and Python in notebooks live under data and AI. This silo keeps the craft of building the software itself.