# Junior software developer

> Turns a described problem into working, maintainable code — and, increasingly, decides whether generated code is actually right.

Page: https://flyvolo.ai/en/careers/junior-software-developer
Impact index: 62/100 · confidence Medium confidence · assessed 2026-09-09
This is not a probability of losing a job.

## Scope

Written specifically for entry-level and early-career software roles, which face a different picture from senior engineering. Six software pages now cut the same work two ways: by seniority (this page and the experienced-engineer page) and by layer (frontend, backend, data engineering). Someone in their first frontend job sits on two of them, deliberately — they answer different questions.

## Tasks

### Writing routine code

- Direction: Automating
- Basis: evidenced (verified records attached)
- Why: Well-specified, heavily represented in training data, and immediately checkable by running it. This is the strongest case for code generation and it happens to be most of what juniors were hired to do.
- What this does NOT establish: Generating it is not the same as owning it. Someone still has to decide it is correct, and that someone has historically learned to judge by writing it first.
- Records:
  - 2026-08-12 · Labour impact · Employment of 22-25-year-olds in the two most AI-exposed occupation quintiles fell about 11% from November 2022 to June 2026, while the least-exposed quintiles grew about 10% — https://flyvolo.ai/en/changes/ev-20260812-junior-software-developer-3 (source: https://digitaleconomy.stanford.edu/app/uploads/2026/08/Canaries_August2026.pdf)
  - 2026-04-13 · Deployment · Thirty-one of forty-five US federal agencies answered yes to generating code using AI in the 2025 federal inventory of commercial AI, some naming GitHub Copilot, Claude Code, Codex and Poolside — https://flyvolo.ai/en/changes/ev-20260413-junior-software-developer-3 (source: https://github.com/ombegov/2025-Federal-Agency-AI-Use-Case-Inventory/blob/3c225ba8438e48306ace7698c8c7feb9486cbc69/Data/2025_consolidated_COTS_AI_use_cases.csv)
  - 2024-10-29 · Deployment · Alphabet's CEO said on the Q3 2024 earnings call that more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers — https://flyvolo.ai/en/changes/ev-20241029-junior-software-developer-1 (source: https://blog.google/company-news/inside-google/message-ceo/alphabet-earnings-q3-2024/)

### Debugging unfamiliar systems

- Direction: Being augmented
- Basis: evidenced (verified records attached)
- Why: Tools are genuinely good at suggesting hypotheses and reading stack traces. They are much weaker at the part that requires holding a model of a specific system in your head and knowing which observation would be decisive.
- What this does NOT establish: This is where the ceiling on generated code shows up most clearly, and it is the skill that separates a junior from someone who can be left alone.
- Records:
  - 2024-04-23 · Deployment · Google enabled ML build repair for all its developers after a randomised trial across half of them; about 34% of developers hitting a broken build in a month apply its suggested fixes — https://flyvolo.ai/en/changes/ev-20240423-junior-software-developer-4 (source: https://research.google/blog/safely-repairing-broken-builds-with-ml/)

### Reviewing generated code

- Direction: New task
- Basis: evidenced (verified records attached)
- Why: Volume of code produced went up sharply; the need to verify it went up with it. Reviewing fluent-but-wrong code is a distinct and newly central skill.
- What this does NOT establish: Reviewing well requires having written the thing being reviewed. If the writing work that trained that judgement is the work being automated, this task has a supply problem within a few years.
- Records:
  - 2024-10-29 · Deployment · Alphabet's CEO said on the Q3 2024 earnings call that more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers — https://flyvolo.ai/en/changes/ev-20241029-junior-software-developer-1 (source: https://blog.google/company-news/inside-google/message-ceo/alphabet-earnings-q3-2024/)

### Turning a vague request into a spec

- Direction: Still human-led
- Basis: inferred (VOLO's reasoning, no attached source)
- Why: Requires context about users, the business and what has already been tried. Generation is downstream of this and amplifies a wrong spec faster than a human would have.
- What this does NOT establish: This is normally a senior responsibility; listing it as a junior task describes where the role is heading, not where most junior jobs are today. The intermediate rungs are what disappeared.
- Records:
  - 2025-07-25 · Pilot · At an IT consulting firm, the analyst estimated LLM-drafted requirements could save 10–15% of drafting time, but they missed user stories whose knowledge was tacit and could not replace the analyst — https://flyvolo.ai/en/changes/ev-20250725-junior-software-developer-6 (source: https://arxiv.org/abs/2507.19113)

### Deciding how pieces fit together

- Direction: Still human-led
- Basis: evidenced (verified records attached)
- Why: Trade-off decisions depend on constraints that live outside the codebase — team size, timeline, what the business will need next. This is what senior means, and it is the destination juniors used to reach by writing routine code for two years.
- What this does NOT establish: Saying seniors are safe is not a statement about entry. The path to senior ran through two years of routine code, and nothing has replaced that path.
- Records:
  - 2026-03-20 · Constraint · The UK Home Office's engineering standard requires every AI-assisted output to be reviewed and approved by a person before production, and keeps accountability for designs with the team — https://flyvolo.ai/en/changes/ev-20260320-junior-software-developer-5 (source: https://engineering.homeoffice.gov.uk/standards/use-ai)

## How to cite this

- The unit is the task, not the job title: say which tasks are changing and which are not.
- Carry each task's basis and its limits with it; an inferred judgement is not a finding.
- This page gives no number of years. Do not derive one from it.
- More: https://flyvolo.ai/llms.txt