← Risk register SOC 15-1252 · reviewed 2026-08-11

Software Developers

1,687,890 US workers · median $135,980/yr · Tech

EXPOSED

Writing functions, unit tests, boilerplate CRUD endpoints, and translating a ticket into a pull request are exactly what coding models do well now, and that is a large share of the median developer's week. What persists is deciding what to build under vague requirements, owning a production system at 3am, negotiating tradeoffs across teams, and being the person accountable when a design choice costs money or leaks data. The occupation splits: the junior implementation tier compresses hard, the systems-and-accountability tier holds, and no license protects either side.

10-year outlook: By 2035 the job title survives but the headcount pyramid inverts — far fewer people paid to type code, more paid to specify, review, and own systems, with entry-level hiring the hardest hit.

US employment, 2021–2025+23.7%
1,364,1801,687,890 workers

Headcount grew steadily across the period.

The job count is not the verdict

This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.

So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing.

BLS projection, 2024–2034

+15.8% 1,693,800 → 1,961,400 on the projections basis

Exposed, but growing

AI can already do a lot of these tasks, and the BLS still expects +15.8% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.

Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.

~115,200 openings a year on average, including replacing people who leave.

One email if this score changes. Watch as many occupations as you like from the same address — no account, and nothing is sent on a schedule, only when a verdict actually moves.

Also known as — 30 job titles this covers

Titles reported by people doing this work, from the US Department of Labor's O*NET survey. If your job title is here, this page is about your work even though the name doesn't match.

DeveloperBeta TesterGame EngineerGame DeveloperServer EngineerSoftware WriterSystem EngineerGraphic EngineerComputer EngineerEmbedded EngineerSoftware DesignerSoftware EngineerSolution EngineerComputer ArchitectInterface DesignerSoftware ArchitectSoftware DeveloperSolution ArchitectUsability EngineerApplication AnalystComputer ConsultantFull Stack EngineerInternet ConsultantSystems Coordinator

Added by hand, not from the survey. O*NET last sampled titles before some of these were in common use, so these are our judgement that the title belongs here — treat them as weaker than the list above. How we decide.

AI EngineerData EngineerDevOps EngineerMachine Learning EngineerPlatform EngineerPrompt Engineer

Score — 41/100 resistance

Holding it up: judgment & accountability (15/20). Weakest point: liability shield (3/20).

Five dimensions, 0–20 each, summed. Higher means more protected. The arithmetic is shown so you can check it: 11 + 3 + 3 + 9 + 15 = 41. · Scored 2026-08-11, and re-examined when evidence accumulates rather than on a schedule.

Task resistance 11/20

Mixed — a routine tier and a judgment tier An 11 reflects the split week: ticket-to-PR work, test scaffolding, API glue, and framework migrations are now draftable in minutes, but the parts that eat the other half of your time — reproducing an intermittent race condition in a distributed system, reading a decade of undocumented business logic to figure out why the invoice total is off by a cent, and deciding which of four broken proposals to ship before quarter-end — still need someone holding the whole system in their head.

Embodiment 3/20

Fully desk- and screen-based A 3 rather than 0 because on-prem deploys, hardware-in-the-loop debugging, and embedded or firmware work put some developers next to physical devices, but the median job is a laptop, an IDE, and a Slack window from anywhere with WiFi.

Liability shield 3/20

No licence, no signature requirement A 3 is where no-licence sits: there is no PE stamp for most software in the US, no bar admission, no statutory sign-off, and CI/CD means a broken commit gets reverted rather than litigated — the employer's EULA disclaims warranty and the company, not you, absorbs the outage.

Trust premium 9/20

Some relationship component A 9 covers what actually holds: users never meet you and your commits are interchangeable to them, but your product manager, on-call rotation, and the two colleagues who know why the payments service is shaped that way rely on accumulated trust and context that a fresh contractor or a model prompt cannot inherit in a sprint.

Judgment & accountability 15/20

Exists to be accountable for ambiguous calls A 15 is earned by the calls with no runbook — approving a schema migration on a live table, choosing to log or not log a field that turns out to be PII, deciding at 3am whether to roll back or forward, and setting an architecture that either scales or gets rewritten in two years — decisions where you are named in the postmortem and the cost is measured in revenue or breached records.

Confidence: high · reviewed 2026-08-11 · how scoring works · 51 deployment reports on file , 1 since this was scored

What this job involves — and which parts are yours

The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.

AI already does these at usable quality

These still need a person

Active moats on the surviving side: judgment

How to future-proof this job

Where to go deeper on what this job runs on: freeCodeCamp — full curriculum, certification at the end free · Coursera — critical thinking and logic, audit free free to audit · Coursera — decision making under uncertainty free to audit · Khan Academy — reading and vocabulary, all levels, free free · MIT OpenCourseWare — full course materials across every department, free free · MIT OpenCourseWare — systems analysis and engineering free

All 35 skills ranked by how many jobs they open →

Where this experience transfers — nothing clears the bar

No occupation passed every test: close enough to software developers on skills and subject matter, at least 10 points more resistant, no big jump in training, no new licence, no pay cut, and not shrinking on its own. That happens for 223 of the 654 occupations here that aren't SAFE, and it is worth stating plainly rather than leaving the section off.

The usual reason is that exposure travels with the skill profile. The jobs most similar to yours tend to be exposed for the same reasons yours is, so the near neighbours don't clear the gap — and the ones that do are a different kind of work, not a transfer of what you already know. Read that as a limit of this method, not a verdict that you're stuck: it only compares whole occupations, and it cannot see specialisation, industry, or anything you'd bring that isn't in a federal skill survey.

Here is that claim on your own job rather than in the abstract. These are the three occupations closest to this one by skill and subject matter — the places the work would most naturally transfer — with what the register scores them:

Database Architects EXPOSED 38/100 (-3) · 85% overlap
Software Quality Assurance Analysts and Testers COOKED 30/100 (-11) · 82% overlap
Database Administrators EXPOSED 38/100 (-3) · 82% overlap

That is the whole problem in three lines. The nearest work is not meaningfully safer, so there is no move here that trades a similar skill set for a better verdict. This is not us running out of ideas — it is what the neighbourhood looks like.

What would move this occupation up is the other direction, and on this page it's the more useful one.

What would move this back up — beyond any one person

The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 55/100, still EXPOSED.

5 specific changes that would raise this score
  • already happening task resistance +4

    Task-mix shift is real here: as ticket-to-PR work is automated, the residual job becomes architecture, incident forensics, cross-service migration under legacy constraints, and reviewing/verifying machine-written diffs at scale. If the median role effectively becomes review-and-integrate of AI output plus system design, measured resistance of the remaining day rises even with no new capability limit — visible already in the collapse of junior openings versus stable senior/staff demand.

  • already happening judgment accountability +2

    Formalized on-call and postmortem accountability tied to regulatory incident reporting — SEC Item 1.05 cybersecurity material-incident disclosure within four business days, and EU NIS2 24-hour reporting — makes the engineer who owns the system the person whose call is documented in a filing. If firms respond by naming system owners of record for critical services, the ambiguity-ownership component hardens.

  • plausible liability shield +4

    Sector-specific attestation regimes that name an individual engineer: the EU Cyber Resilience Act (obligations phasing to 2027) plus CISA's Secure Software Development Attestation Form, which requires a named company officer to attest to SSDF practices for software sold to the US federal government. If attestation devolves to a named responsible engineer with personal exposure — or if a licensure scheme like the (currently withdrawn in most states) PE Software Engineering exam is revived for safety-critical systems such as medical device firmware, avionics, or voting systems — a signature requirement appears where none exists.

  • plausible liability shield +2

    FDA Software as a Medical Device and DO-178C/FAA airworthiness pathways already require named individuals to sign verification records for a narrow slice of developers. If FDA guidance on AI-enabled device software function requires human-identified authorship and sign-off on model-generated code changes in premarket submissions, that slice widens.

  • plausible trust premium +2

    Narrow route only: procurement and insurance language demanding provenance. Cyber insurers and enterprise vendor questionnaires increasingly ask whether AI-generated code entered the codebase, and some contracts (and license-contamination fears after the GitHub Copilot litigation) require warranty that code is human-authored or human-reviewed with named reviewers. That is willingness to pay for human authorship, but it is contractual rather than consumer taste, and it caps low.

The limit. No realistic route to a broad consumer trust premium — buyers of software want the artifact, not a human's hand in it, and no client asks who typed the function. Embodiment has no route. Even with attestation regimes, licensure would cover a small safety-critical minority; the bulk of the occupation remains unshielded, and the compressing junior tier gains nothing from any of these levers.

These are conditions, not forecasts — what would have to happen, not what will. Specific rules, cases and bills are named so you can go and check whether they exist and where they stand; verify before relying on any of them. Nothing here is legal or financial advice.

Where this work is, and what it pays there

BLS metro figures for 382 areas. The verdict above does not change by city — the rubric judges what the work involves, not where it happens — but pay and headcount do, and the national median hides a very wide range.

Most of these jobs

New York-Newark-Jersey City, NY-NJ 121,000 $166,830 +23%
Seattle-Tacoma-Bellevue, WA 92,770 $167,280 +23%
San Jose-Sunnyvale-Santa Clara, CA 87,350 $213,110 +57%
Washington-Arlington-Alexandria, DC-VA-MD-WV 69,060 $154,930 +14%
San Francisco-Oakland-Fremont, CA 69,030 $186,640 +37%
Dallas-Fort Worth-Arlington, TX 67,030 $133,290 -2%
Los Angeles-Long Beach-Anaheim, CA 55,540 $160,920 +18%
Boston-Cambridge-Newton, MA-NH 42,310 $166,090 +22%

Best paid

San Jose-Sunnyvale-Santa Clara, CA 87,350 $213,110 +57%
San Francisco-Oakland-Fremont, CA 69,030 $186,640 +37%
Seattle-Tacoma-Bellevue, WA 92,770 $167,280 +23%

Percentages are against this occupation's national median of $135,980. Counts are jobs in that metro, not vacancies. Metros where the BLS suppressed the cell are absent rather than shown as zero.

Who is actually doing this

The score above is about what the work exposes. This is reporting about real deployments in this occupation — the difference between "could be automated" and "somebody automated it."

Meta · Oracle · Amazon · Block · Monday.com · DeepL · Salesforce · National University of Singapore · Rapid7 · Microsoft · Visa · Electronic Arts · Life360 · Mews · City of Charlotte, North Carolina · Take-Two Interactive · UBS · Magnific · Maya · Globant · Coinbase · PwC · Cisco; Block · Meta, Microsoft · Snap · IBM France · HP

6 of 105 reported cases, with sources

99 more in the dispatch

Quick take — do you do this job?

Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.

Self-reported and unverified — a sentiment signal, not a survey. One response per person per occupation; you can change your answer.

Field reports — what people say has changed

No field reports yet. A written account takes a paragraph rather than a tap, goes to an editor before it appears, and is the one thing on this page the rubric cannot produce on its own.

File a field report

Concrete beats general: a tool that arrived, a task that moved, a headcount decision you watched happen. Don't include anything that identifies you or your employer if that would put you at risk.

Watch this verdict
Kept current

Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.