The Job Didn't Disappear. It Moved Up the Stack.

The fear has been simple: if models write the code, why hire developers? The recent labor data says something more interesting. The job is not vanishing. It is concentrating in people who can direct, evaluate, and integrate agent work.

What the postings show#

CNBC reported on October 11, 2026 that Indeed Hiring Lab found U.S. software-development job postings up nearly 15% since late February 2025, even while overall job postings fell about 7%. That rebound still left postings roughly 27.5% below their pre-pandemic peak, so this is not a return to the old boom. It is a selective recovery.

The composition matters more than the headline. Between May 2025 and May 2026, senior roles accounted for 71% of the increase in software-development postings. Job titles that mention AI accounted for 37% of the increase. Companies appear to be hiring people who can oversee AI-generated work, refine products, and connect tools into broader systems.

GitHub's own volume numbers sit beside that hiring story. The same CNBC piece noted GitHub's report that one in three pull requests now involve an AI agent, up from fewer than one in ten a year earlier. More code is moving. The scarce skill is deciding what should move and whether it is safe.

October kept the same shape#

CompTIA's November 6, 2026 analysis of October data pointed the same direction. Tech industry employment rose by about 7,200 jobs. IT occupations rose by about 26,000. Employers added 290,892 new IT occupation postings in October, up 4.5% from September and 22.2% from October 2025.

Software developer and engineer roles led with 51,944 new postings, up 6.3% month over month. Active postings associated with AI skills reached 366,322, up 6.2% from September. Specialized AI positions rose 9.8%. Demand is showing up both in AI-titled jobs and in ordinary roles that now expect AI capability.

Who is under pressure#

The hard edge of the story is early career. Stanford's 2026 AI Index, cited by CNBC, found employment in AI-exposed fields for ages 22 to 25 down 16% since the rise of generative AI, with early-career software developers seeing about a 20% decline even as older developer roles expanded.

That should worry anyone who leads a team. Today's junior pipeline is tomorrow's senior bench. If companies only hire people who already know how to supervise agents, they will eventually run out of people who learned judgment the long way.

The Bureau of Labor Statistics still projects software developer employment to grow about 10% from 2025 to 2035, from roughly 1.72 million to 1.89 million jobs. The long-term demand is there. The apprenticeship path is the open question.

How I read it as a leader#

I would not hire for "can write code without help." I would hire for a shorter list:

1. Can break a problem into work an agent can attempt.
2. Can tell good output from confident nonsense.
3. Can connect a change to product, data, security, and operations.
4. Can teach the system what failed so the next run is better.
5. Can communicate clearly with people who do not live in the codebase.

Those skills used to be what seniors grew into after years of implementation. Now they are the job description earlier. That does not mean juniors are useless. It means teams have to design junior work around review, evaluation, integration, and owned slices of a system, not around endless greenfield typing that an agent can draft overnight.

The quiet management task#

Agent tooling will keep getting better. Postings will keep oscillating. The management task underneath is steadier: protect quality while volume rises, and protect learning paths while senior demand dominates.

If your hiring plan only chases AI-titled seniors, you are buying capacity for this quarter. If you also build a path for people to learn judgment beside agents, you are buying a team that still works in five years.