Only 1 in 10 tech engineering jobs is entry-level

Across 9,338 tech engineering listings posted to Greenhouse, Lever, Ashby and other major ATSes between February 28 and June 29, 2026, 11.0% are open to entry-level engineers. 54.9% target senior, staff, or principal levels. Senior-and-above listings outnumber entry-level ones by roughly five to one.

The pattern is sharpest where the on-ramps used to be. DevOps engineering, once a refuge for self-taught generalists, posts 13 senior listings for every entry-level one. Backend engineering, the classic new-grad entry point, posts nine senior for every entry-level. Remote listings, which once helped entry-level engineers escape expensive cities, are 8.6 to 1; onsite roles are 3.3 to 1.

Of 224 companies posting 10 or more tech engineering roles in our window, 63 posted zero entry-level listings. GitLab, Grafana Labs, Square, ClickUp, Replit, Dropbox, and Perplexity all sit in that group.

The door into tech engineering is still open. It is mostly open to engineers with at least four years of experience.

The headline number

Of the 9,338 tech engineering listings posted to supported ATSes since February 28, 2026, 1,028 (11.0%) target entry-level engineers. 3,006 (32.2%) target mid-level. 5,125 (54.9%) target senior, staff, or principal. The remaining 1.9% are director and above.

Senior + Staff + Principal

54.9%

Mid

32.2%

Entry-level

11.0%

Director and above

1.9%

For context, Indeed's Hiring Lab reported in late 2025 that entry-level tech postings had fallen from 8.1% to 7.4% of the IT job mix year over year, while senior-level postings climbed from 38.8% to 43.1%. Our number is higher because our scope is narrower (tech engineering specifically, not all of IT). The shape of the imbalance matches almost exactly.

The senior shift is not new. Indeed reported in mid-2025 that the share of tech postings asking for at least five years of experience rose from 37% in Q2 2022 to 42% in Q2 2025. Stanford's Brynjolfsson, Chandar, and Chen documented a 13% relative employment decline for early-career workers in the most AI-exposed occupations since the rollout of generative AI, with software development among the hardest-hit groups. Our snapshot is what that decline looks like from the listing side of the market.

The senior-to-entry-level ratio across the data is 4.99. We round it to five to one in prose because the rounding does not change anything.

What we measured

The dataset is every job listing in the Hotfix index with a posted_at between February 28 and June 29, 2026 in 14 tech engineering functions: backend, frontend, fullstack, devops, machine learning, data engineering, data science, security, hardware, embedded, support, sales, GTM, and solutions architecture engineering. February 28 is the start of our posted_at tracking window. June 29 is the query cutoff.

Excluded functions: engineering management, technical program management, and IT support. These exist in our normalization but are managerial or auxiliary; they are not the on-ramps an entry-level engineer typically targets.

For each listing we use Hotfix's normalized seniority_level field, inferred at ingestion from title plus description signals (years-of-experience requirements, level codes like SDE I or L4, explicit junior or senior markers in the title, internship and new-grad markers). Categories: entry-level, mid, senior, staff, principal, director, vp, executive.

In this analysis "entry-level" is the bottom IC rung (1,028 listings, 11.0%) and includes listings whose title or description signals 0-2 years of experience, internship, new-grad, or associate status. "Senior and above" is the union of senior, staff, and principal (5,125 listings, 54.9%). Mid is its own category at 32.2%. Director and above are presented separately at 1.9%; they are individual-contributor leadership listings that distort the seniority picture if mixed in.

The dataset behind this analysis is available as a CSV. (Casey: link goes here.) It includes per-listing fields, function assignments, and the named-employer table.

The most foundational functions are the most closed

The functions that have historically employed the most engineers are the most senior-stacked.

Support engineering

28.9%

Embedded engineering

19.7%

Fullstack engineering

17.5%

Hardware engineering

16.8%

Solutions architecture

11.1%

ML engineering

10.5%

Frontend engineering

10.0%

Sales engineering

9.9%

Data science

9.8%

Data engineering

8.9%

Backend engineering

7.8%

GTM engineering

7.8%

Security engineering

7.6%

DevOps engineering

5.0%

DevOps engineering posts 13 senior, staff, or principal listings for every entry-level one. Backend engineering, 9 to 1. Security engineering, 7 to 1. Data engineering, 7 to 1. These are the functions that built most early-career careers a decade ago.

The pattern weakens at the edges. Support engineering is the only tech engineering function in our data where entry-level listings outnumber senior ones: 28.9% entry-level, 19.2% senior-and-above. Embedded engineering, often associated with hardware research and apprenticeship culture, is 19.7% entry-level. Hardware engineering, despite being the function with the worst ghost-job rate in our Q2 2026 ghost-job study, is 16.8% entry-level. Fullstack engineering, the catch-all that smaller startups post when they want generalists, is 17.5% entry-level.

The implication is uncomfortable. The functions that built most early-career careers (backend, devops, security) are now the functions most closed to early-career engineers. The functions that remain meaningfully open are the ones that have always existed at the edges of capital-E Engineering: support, embedded, and small-shop fullstack.

ML engineering deserves a note. It is 10.5% entry-level, higher than backend (7.8%), data engineering (8.9%), or data science (9.8%). The simple narrative that AI roles are absorbing the engineering pool while pushing out juniors does not match our data. ML roles are slightly more open to entry-level engineers than the foundational eng roles they are supposedly replacing. The bottleneck is not at the AI end; it is at the foundational end. AI did not eat the entry-level backend engineer. The senior backend engineer did.

Remote is closed to entry-level engineers

Remote tech engineering listings are the least open to entry-level engineers. Only 6.9% of remote tech eng listings target entry-level, compared to 11.4% of hybrid and 15.1% of onsite.

Onsite

15.1%

Hybrid

11.4%

Remote

6.9%

The senior-to-entry-level ratio: 8.6 remote, 4.8 hybrid, 3.3 onsite. A remote tech engineering listing is more than two and a half times as senior-stacked as an onsite one.

The mechanism is not mysterious. Remote hiring is operationally expensive at the entry-level end. A new engineer hired remote needs onboarding scaffolding that few companies have built: a working mentor at the other end of a video call, a culture of asynchronous documentation, deliberate pair-programming time. Senior remote hires self-onboard. Companies that need to ship default to senior remote and entry-level onsite, because the second is the only configuration that works.

This is one of the quieter consequences of the post-pandemic remote-work compression. The remote-first pitch was sold to early-career engineers as a way out of expensive cities and into the global talent pool. In practice, the companies that built durable remote cultures use them to hire experienced engineers from anywhere, while early-career engineers who want their first job have been pushed back into commuting distance.

It also helps explain a finding from our Q2 2026 ghost-job study: onsite tech listings ghost at nearly twice the rate of remote (44.9% vs 26.5% at 60 days). One reading is that onsite is where companies park their lower-conviction, more-junior reqs, and lower-conviction reqs ghost.

Twenty-eight percent of high-volume tech employers posted zero entry-level listings

Of the 224 companies posting 10 or more tech engineering listings in our window, 63 (28.1%) posted zero entry-level listings. Not one.

Some of those companies are unsurprising. Late-stage developer tools and infrastructure businesses (GitLab, Grafana Labs, Clickhouse, Drata, Square) have always staffed engineering heavily with experienced ICs. Some are more interesting. Replit, a company whose product encourages first-time programmers to learn by building, posted 31 tech engineering listings in our window and zero at the entry-level tier. Perplexity, an AI search company aggressively hiring through 2025-2026, posted 27 and zero. Deepgram, 26 and zero. Turion Space, 26 and zero.

The full list is in the CSV companion.

A few caveats. Our job-function scope excludes engineering managers, TPMs, and IT support, so a company that hires entry-level engineers only into those tracks would still appear here as zero. Our seniority normalization can mis-tag specific listings (a "Software Engineer" with no level signal usually defaults to mid). We do not see internships or rotational programs that never appear on the public ATS feeds. With those caveats noted: zero is still zero across 26 to 54 listings per company. A company posting 39 tech engineering listings and tagging none of them entry-level is making a statement about who it wants to hire.

The current floor

The imbalance is not getting worse within our window. Across three six-week slices, the entry-level share holds nearly flat.

10.0%

12.2%

10.3%

Feb 28 - Apr 14

Apr 15 - May 30

May 31 - Jun 29

The 11% headline is not a sliding number that will be 9% by Q4. It is the current floor. Companies have settled at this seniority mix and are holding.

That matters for the macro narrative. The AI-killed-entry-level engineering story tends to be told as a continuous acceleration: every quarter worse than the last. Our four months suggest the acceleration phase, whenever it happened, is over. What we are looking at is a steady-state market in which hiring at the entry-level tier has been deliberately downsized and is now holding at the new level.

In some ways that is the worse interpretation. A still-falling number eventually bottoms. A floor stays.

What this means

For new and early-career engineers, the listing-side picture matches the lived-experience reports of the past two years. Applications go into Greenhouse, Lever, and Ashby. They come back as rejections or silence. The structural reason is visible in our data: about nine of every ten tech engineering listings on those systems are not, in fact, for you.

Given this floor, four levers actually move:

Target the functions that remain open. Support engineering, embedded engineering, fullstack at smaller companies, and ML engineering are all above the 10% line, and several are well above it. Backend and devops are the worst on-ramps in our data.

Target onsite work. The 6.9% entry-level share on remote listings is the most senior-stacked work configuration in our data. Onsite at 15.1% is more than twice as junior-friendly. This is unwelcome news to anyone who was sold remote-first as a junior-friendly future.

Target smaller companies. Companies of 1-50 people are 16.4% entry-level, against 9.9% at companies of 201-1000 and 10.0% at companies of 1000+. The biggest companies are the most senior-stacked. They always have been, but the gap has widened, and the new-grad packages the NACE 2026 Job Outlook celebrates (CS starting salaries up nearly 7% to roughly $81,535) are concentrated at exactly the wrong end of that distribution.

Treat the zero-entry-level list as a do-not-spend-cycles list. If GitLab has posted 54 tech engineering listings since February and zero at the entry-level tier, that is signal about GitLab's current hiring posture. Apply if you have the experience. If you are graduating in May, look elsewhere.

For employers, the data raises an awkward question: who will be your senior engineers in 2030? The pipeline that produced senior engineers in 2025 ran through entry-level hiring in 2018-2020. The pipeline that produces senior engineers in 2030 has to run through entry-level hiring now. Our data shows that pipeline being deliberately starved. The cost is not visible yet. It will be.

Lastly, for the press: please stop writing "entry-level tech jobs are disappearing" without saying how many. It is 11%. Specifically, in tech engineering on the major ATSes, since February 2026, it is 11%. The number is not zero. It is also not what it used to be.

Counterpoints

Three.

Entry-level hiring happens at career fairs, not on ATSes. This was true a decade ago. It is no longer true. Most full-time new-grad hiring at companies of any meaningful size now flows through the same ATSes we ingest. Stripe, Google, Meta, Microsoft, and the rest of the household tech employers post new-grad reqs on Greenhouse or equivalents. We do undercount internships (which often close before public posting), rotational programs, and university partnership pipelines. We do not undercount full-time new-grad engineering hiring at companies with public careers pages.

One quarter does not prove a trend. We are not claiming the trend in this post. We are claiming the current floor. The historical drop comes from external sources, notably Indeed Hiring Lab (entry-level tech postings down from 8.1% to 7.4% of the IT job mix year over year) and Stanford's Brynjolfsson-Chandar-Chen analysis of ADP payroll data (13% relative employment decline for early-career workers in AI-exposed occupations). Our data anchors the present.

Inferred seniority is noisy. It is. A "Software Engineer II" or "Software Engineer" with no level signal usually lands in mid, when in some companies' systems it arguably means entry-level. We do not claim listing-level accuracy. The bucket aggregates are robust to plausible misclassification. The qualitative findings hold across reasonable variations of the threshold: foundational eng is more senior-stacked than fringe eng; remote is more senior than onsite; large companies are more senior than small. Under a different normalization, the exact decimal of 11.0% would land between roughly 9.5% and 12.5%; the shape does not move.