Executive summary
Only 6.8% of 8,618 classified AI/ML engineering roles are open to juniors (583 of 8,618). For every entry-level opening there are 9.8 senior-plus roles (5,707 senior + lead + principal). This is the tightest junior-to-senior ratio in any tech specialty tracked by the AI Dev Jobs index. Breaking in is statistically rarer than staying in.
The squeeze: experience-level mix
Across the 8,618 AI/ML engineering postings with an experience-level classification, the distribution is top-heavy. Junior is the smallest band; senior, lead, and principal together account for 66%.
| Experience level | Role count | Share |
|---|---|---|
| Junior / Entry-level | 583 | 6.8% |
| Mid | 2,328 | 27.0% |
| Senior | 2,916 | 33.8% |
| Lead / Staff | 2,270 | 26.3% |
| Principal | 521 | 6.0% |
The junior band is smaller than the principal band in absolute terms on some crawl cycles, even though there are orders of magnitude more eligible junior candidates. This isn't a supply problem — it's a demand signal. Companies are spending budget on top-of-stack, not bottom-of-stack, AI talent.
What "junior AI" actually pays
The public stats endpoint does not expose salary by experience band directly — it publishes a single global average ($224,911) and a global median ($213,175) across the 3,402-posting salary-disclosed subset. But we can triangulate: of the disclosed postings, 367 advertise a midpoint under $150k (a reasonable proxy for the entry-level band), while 457 advertise $300k+. If junior supply is 7% of roles, the under-$150k band is roughly at odds with that share — meaning advertised junior AI pay typically lands in the $100k-$150k range, well above non-AI entry-level engineering but well below the $213,175 index median. Data gap flag: the next iteration of this report will break salary by experience band once the /api/v1/stats endpoint exposes it.
Why the squeeze exists
Three structural reasons explain the junior squeeze. Judgment under uncertainty. AI/ML engineering outputs (prompts, evals, RAG retrievers, agent pipelines) don't have deterministic unit tests — they require a senior's pattern-matching to spot when a system is quietly wrong. Companies willing to pay $224,911 average don't want to train that judgment from scratch. Compute-cost economics. A senior engineer ships iterations against GPU-hours that cost $5-$50 per experiment. The efficiency delta between a senior and a junior running GPU jobs compounds faster than it does in CRUD web development, so staffing leans toward senior. The ML-plus-LLM double requirement. Companies hiring for "AI" in 2026 typically want traditional ML fluency (PyTorch, distributed training, MLOps — the 2,297 postings combining those skills) plus recent LLM / agent experience. Juniors rarely have both simultaneously.
Bootcamps, CS grads, and career-switchers
Is the bootcamp path viable? The data says: narrow pipe, not closed pipe. The junior AI/ML market is roughly 583 roles — a fraction of the 8,618-role total, but still 583 concrete opportunities across 513 companies in the index. The realistic read for a career-switcher: (1) the generic "AI engineer" bootcamp targeting LLMs alone will not beat an ML-native CS grad on the same req, (2) the working angle is a domain wedge — AI-for-robotics, AI-for-healthcare, AI-for-security, AI-for-legal — where the ML stack is a smaller share of the role than the domain knowledge, (3) research labs (Anthropic, OpenAI, DeepMind) post "Resident" and "Fellow" tracks that function as senior-paid junior seats. These look like the highest-leverage entry points in the current market. The 7% share is a filter, not a ceiling.
Companies that DO hire juniors
The 6 companies below showed at least 3 junior-titled AI/ML roles in the sampled paginated index walk. Titles counted as junior match junior, jr, entry-level, new grad, associate, intern, or early career. These are the companies most worth watching if you're breaking in — frontier labs and large AI-first companies still post entry-level seats, even when the aggregate mix is 7% junior.
| Company | AI/ML roles (sample) | Junior titles | Junior share |
|---|---|---|---|
| Torc Robotics | 19 | 7 | 37% |
| 24 | 7 | 29% | |
| Datadog | 12 | 5 | 42% |
| Lila Sciences | 24 | 5 | 21% |
| Carbon Robotics | 11 | 3 | 27% |
| Waymo | 73 | 3 | 4% |
Signal for job seekers
If you're trying to break in. Don't optimize for the 513-company index overall — the signal is too diluted. Target the 6 of companies actually posting junior seats (see table above). Apply to research-lab residency programs that pay senior-band salaries to juniors. Specialize in a non-trivial domain (robotics, biotech, finance, defense) before the AI layer — AI-for-X always hires junior talent faster than pure-ML orgs do, because the domain expertise is the harder-to-hire input.
Methodology
Generated from the live aidevboard.com/api/v1/stats endpoint (public, unauthenticated) plus a paginated walk of /api/v1/jobs for per-company experience-level aggregation. Index scrapes 560+ ATS sources on a daily cron and deduplicates by (company, title, location). The experience_level field is classified by a rules-based parser from title + seniority signals — see open-source parser. Per-company junior counts in this paper use a title-regex fallback (junior, jr, entry-level, new grad, associate, intern, early career) over the paginated job sample — they're a proxy, not a source of truth, and skew conservative (any title with "senior" anywhere is dropped from the junior bucket). Of 8,618 roles in the index, 8,618 have a non-empty experience-level classification. Salary averages are from the 3,402-posting salary-disclosed subset. This page auto-regenerates weekly (Mon 9:15 am PT).
Download raw data: The experience-level + per-company junior dataset is mirrored as a public gist — CSV · Markdown · view on GitHub. Auto-updated every weekly regeneration; canonical raw URLs are stable across revisions.
What's next
For the top-line AI hiring landscape, see Q2 2026 AI Engineering Hiring Snapshot. For compensation across skill tags, see Q2 2026 AI Compensation by Skill. For the workplace-pay angle (remote vs onsite vs hybrid), see Q2 2026 Remote vs Onsite AI Hiring. For the agent-infra side of the market, see Q2 2026 MCP Ecosystem Health. Full reading paths at the Research Atlas.