The gap in numbers: a market that cannot fill its own demand
As of June 2026, the global AI labour market is running a structural deficit that no short-term hiring campaign will resolve. Research from Second Talent, published via PRNewsWire, puts total open AI positions at approximately 1.6 million worldwide against a qualified candidate pool of roughly 518,000 — a 3.2:1 demand-to-supply ratio. For context, software engineering as a whole has historically sat closer to 1.4:1 in peak years.
The shortage is not evenly distributed across disciplines. Roles requiring deep applied experience in generative systems, autonomous robotics, and AI governance are seeing the steepest gaps. The table below shows the shortfall percentage by role type, based on current open requisitions versus available candidates.
| Role type | Open positions (est.) | Shortage % | Trend (YoY) |
|---|---|---|---|
| Robotics & Automation Engineer | ~290,000 | 70% | Worsening |
| Generative AI Engineer | ~380,000 | 67% | Worsening |
| AI Governance & Policy Specialist | ~85,000 | 52% | Emerging |
| ML Infrastructure Engineer | ~210,000 | 48% | Stable |
| Applied AI Researcher | ~175,000 | 44% | Stable |
| AI Product Manager | ~130,000 | 39% | Growing |
What these figures obscure is the entry-level dimension. When you filter to roles targeting practitioners with fewer than three years of experience, the picture becomes considerably more complicated — and considerably more interesting for anyone who is early in their career right now. More on that shortly.
For a broader view of how these numbers have shaped compensation expectations, see our AI engineer salary data and AI skills shortage coverage from earlier this year.
India: a market growing faster than it can train
India's AI job market is expanding at a pace that its own talent pipeline cannot match. As of June 2026, 11.7% of all job postings in India explicitly require AI skills (per igmGuru/taggd.in salary research, June 2026) — up from 8.2% the previous year. That is a 43% jump in the share of postings in just twelve months. Demand is growing at roughly 40% year-on-year while the qualified talent pool is expanding at only 15–20%. The compounding shortfall is visible in hiring timelines: engineering teams report average time-to-fill for senior AI roles of 4–6 months, up from under 90 days two years ago.
The funding environment is amplifying this further. India attracted $1.48 billion in AI startup funding in Q1 2026 alone — 38% of total startup funding for the quarter. Neysa's $1.2 billion Series B, India's largest single AI funding round to date, is the headline number, but the more instructive story is in the mid-market: dozens of Series A and B companies are simultaneously competing for the same shallow pool of AI-skilled engineers.
For practitioners entering or early in their careers, this translates directly into salary leverage. Entry-level compensation in India has shifted materially over the past 18 months.
| Role | India salary range (2026) | Notes |
|---|---|---|
| AI Engineer (fresher, 0–1 yr) | ₹5–9 LPA | Varies by city; Bengaluru/Mumbai top of range |
| GenAI Specialist (1–2 yr) | ₹8–12 LPA | Strong demand from product startups |
| ML Infrastructure Engineer (2–4 yr) | ₹12–20 LPA | Highest competition for candidates |
| Applied AI Researcher (2–4 yr) | ₹14–22 LPA | IIT/research-institution premium visible |
| Top earner (IIT/FAANG-adjacent) | ₹15 LPA+ | Outliers at funded unicorns exceed ₹30 LPA |
Salary growth is averaging 15–20% annually and that trajectory is expected to hold through 2030, assuming continued investment and no major correction in AI funding. The transition roadmap for software engineers considering an AI pivot covers exactly which skills unlock the biggest step-changes in compensation.
UK: serious money, serious shortage
The United Kingdom's AI sector raised £6 billion in 2025 — an 80% increase from 2024. Q1 2026 alone accounted for £3 billion of that, suggesting the pace is not slowing. This is not abstract macroeconomic data: it is a direct signal that UK companies are capitalising, hiring, and competing for talent they cannot find through conventional channels.
Entry-level AI salaries in the UK are notably high by European standards. Practitioners with under one year of experience are commanding £54,810–£70,734 at advertised rates (per Glassdoor and Machine Learning Jobs UK, June 2026), with specialist premiums above that floor for anyone who can demonstrate hands-on generative AI or ML infrastructure experience. The table below reflects current advertised ranges rather than theoretical maximums.
| Role | UK salary range (2026) | Notes |
|---|---|---|
| AI Engineer (<1 yr experience) | £54,810–£70,734 | London commands ~12% premium over national rates |
| GenAI Developer (1–2 yr) | £65,000–£85,000 | High demand from fintech and healthtech |
| ML Engineer (2–4 yr) | £80,000–£110,000 | Severe shortage; strong counter-offer activity |
| AI Governance Lead (3–5 yr) | £75,000–£100,000 | Emerging category, very few qualified candidates |
| AI Infrastructure Specialist (3+ yr) | £95,000–£140,000 | Competitive with US remote offers |
The UK's regulatory environment — with the FCA, Bank of England, and ICO all accelerating AI guidance — is creating a secondary demand wave for AI Governance specialists that did not meaningfully exist two years ago. This is a category where early movers with even modest formal experience have outsized leverage.
For more on how UK-specific hiring dynamics are evolving, our AI hiring trends coverage goes deeper into retention challenges and the counter-offer environment.
The paradox: entry-level postings down 35%, demand up
Entry-level AI job postings are down 35% since 2023 — but demand for AI-skilled workers at junior salaries is rising. This is not a contradiction; it is a structural shift in how companies hire. Understanding it is the difference between a frustrating job search and a short one.
Here is what happened. Between 2022 and 2024, the technology industry went through significant workforce reductions. Many companies eliminated entry-level AI requisitions rather than filling them, consolidating work upward. When hiring resumed, surviving job descriptions were rewritten with inflated experience requirements — 4–6 years became the baseline for roles that previously targeted 1–2 years.
The result is a documented absurdity: only 2.5% of AI engineering roles today explicitly target practitioners with 0–2 years of experience, despite the fact that demand for AI-capable early-career workers has increased year-on-year. Employers want AI skills at junior salaries but have not updated their job descriptions to reflect what they are actually willing to hire.
There is a second structural shift compounding this. As of June 2026, 28% of entry-level AI positions are filled before they appear on public job boards (per Zero To Mastery's 2026 hiring research). The implication is direct: the conventional job-search funnel — find posting, submit CV, wait — systematically excludes a significant fraction of available opportunities. The roles are being filled through networks, referrals, and direct outreach to visible practitioners before the requisition ever goes live.
The 28% figure is not a fluke — it reflects how hiring managers at well-funded teams actually operate. They shortlist from people they know or can find online before the requisition is formally opened. If you are not publicly discoverable with a clear signal of what you build, you are invisible to this channel entirely. A verified profile on a directory that hiring teams actually browse is not a nice-to-have; for early-career practitioners, it is access to a quarter of the market that the traditional job board will never show you.
What funded companies are actually hiring for
Across both the Indian and UK markets, the skills signal in live job postings — and in the roles being filled through networks before posting — clusters around a set of capabilities that are relatively recent and still undersupplied in the candidate market. The following skills appear most frequently in AI requisitions from venture-backed companies as of Q2 2026.
- Retrieval-augmented generation (RAG) pipelines — implementation, evaluation, and production optimisation. Almost every funded product company needs this and very few early-career candidates can demonstrate it with shipped work.
- Fine-tuning and instruction-tuning on open-weight models — Llama 4, Mistral, and smaller domain-specific variants. Experience with PEFT, LoRA, and QLoRA is specifically valued.
- LLM evaluation and red-teaming — the ability to build systematic evaluation harnesses, not just prompt and observe. Companies that have shipped once are investing heavily here.
- Inference infrastructure and cost optimisation — batching, quantisation, speculative decoding, and serving with vLLM or equivalent. ML infra engineers who can cut inference cost by 30% are in extremely short supply.
- Agent frameworks and tool-use patterns — multi-agent orchestration, MCP integrations, and production agent observability. This is the fastest-moving area and the skills gap is widest here.
- AI governance and responsible AI — risk assessments, model cards, bias audits, and regulatory mapping (EU AI Act, UK Frontier AI Bill, India's DPDP Act). AI governance is a 52% shortage category and almost no early-career practitioners are explicitly targeting it.
- Vector databases and embedding pipelines — Pinecone, Weaviate, Qdrant, and pgvector. Practical experience building and maintaining production vector stores is still rare at the junior level.
The pattern across these skills is consistent: they are all learnable in 3–6 months of focused project work, they are all verifiable through public portfolio evidence, and they are all currently undersupplied in the candidate market. The portfolio guide covers how to package this evidence for maximum hiring signal.
Why being findable beats cold applying
The orthodox job-search advice — tailor your CV, write a cover letter, apply on LinkedIn — is not wrong, but it addresses only the fraction of the market that operates through formal postings. Given that 28% of entry-level AI roles are filled before appearing on any board, and given that hiring managers at funded companies are actively shortlisting from discoverable practitioners, visibility is a distinct and complementary strategy to application volume.
Consider what a hiring manager at a Series B AI infrastructure company actually does when a requisition is approved. They open a spreadsheet of people they have heard of or can quickly verify. They search their network. They check directories and communities. They look at who has been writing or speaking about the relevant technical problem. Only if those channels fail — or if they are hiring at scale — does the formal posting go live.
"I had applied to about forty roles through the usual job boards with no traction. Two weeks after adding my AI Tech Connect profile, I had three inbound messages — two from UK fintech teams I had never heard of and one from a Bengaluru GenAI startup that had seen my RAG project write-up. I ended up interviewing for all three. The difference was not my CV; it was being findable in a place where people who were actually hiring were looking."
— Kavya, Verified Builder · Bengaluru, IN / London, UK (illustrative)This is not anecdotal noise. The structural shift toward private hiring — driven by the reduction in formal entry-level postings, the rise of network-based shortlisting, and the speed at which funded companies want to move — means that discoverability is now a first-class career asset. It does not replace skills or demonstrated work; it amplifies them by ensuring they reach the people who need them.
The discoverability argument also compounds over time in a way that cold applications do not. A profile that signals a specific technical capability — say, production RAG pipelines with LlamaIndex — continues to surface in searches long after it is created. Each article, project, or contribution linked from that profile adds signal. Cold applications have a shelf life of roughly 72 hours before they are buried.
1.6 million open roles. Are the right teams finding you?
AI Tech Connect is where Indian and UK AI builders get verified and get found. Browse Verified Builder profiles or add yours in under ten minutes.
Claim your profile →Your AI Tech Connect profile as the fix
AI Tech Connect exists precisely at the intersection of the two problems this article has described: too many unfilled roles, and too many capable practitioners who are invisible to the teams filling them.
A Verified Builder profile on AI Tech Connect is not a CV copy-paste. It is a structured signal of what you build, what stacks you ship on, and where you are. It is indexed by hiring teams across India and the UK who use the directory specifically because they are looking for verified practitioners — not anonymous applications. The shortlist-five-and-email model means that hiring managers can reach out directly to practitioners they want to speak with, cutting the formal requisition step entirely.
For practitioners with 0–2 years of experience, the profile is particularly high-leverage. Given that only 2.5% of AI engineering roles formally target this experience bracket, your profile needs to do work that a job board application cannot: it needs to surface you in contexts where experience filters are not applied — inbound discovery, referral, and shortlisting before a role is formally posted.
The Founding Builder programme is the time-sensitive element. There are fewer than 500 Founding Builder spots available. Practitioners who claim a profile during this window receive a permanent Founding Builder badge — a visible, verifiable signal that they were among the first cohort on the platform. As the directory grows, that badge carries increasing weight as a proof-of-early-commitment signal. It is the kind of credential that costs nothing and compounds indefinitely.
Getting started takes under ten minutes. Add your name, your current stack, your two or three strongest projects with links to shipped work, and a brief statement of what you are building or looking to build next. That is enough to surface in hiring searches and start generating inbound. For more detail on what to include, the portfolio guide covers the evidence structure that converts profile views into conversations.
The gap is real, the numbers are large, and the window is open. The question is whether you are positioned to capture it.
What to do this week
The data in this article describes a structural opportunity, but structure does not translate into outcomes automatically. Here is a concrete sequence for early-career AI practitioners in India or the UK who want to act on what the gap represents.
- Audit your discoverability this week. Search your name on Google alongside "AI engineer" or your primary stack. If nothing useful surfaces in the first page, you are invisible to the 28% of roles filled before posting. Fix that first.
- Add your AI Tech Connect profile. Go to /submit/. Spend ten minutes on it. Link your two strongest shipped projects. If you do not have shipped projects yet, the portfolio guide will tell you what to build and how to document it.
- Pick one undersupplied skill from the list above and spend the next 30 days building something demonstrable with it. A working RAG pipeline, a fine-tuned model, a production evaluation harness — anything with public evidence. Attach it to your profile.
- Browse the directory. Look at Verified Builder profiles from practitioners a year or two ahead of you. See how they describe their work. Use it as a calibration for how to frame your own.
- Read the tips and guides section. The transition roadmap is specifically written for practitioners making the move from general software engineering to AI roles. It covers the skills, the timeline, and the signal that actually moves hiring conversations forward.
The talent gap will not last for ever. The cohort of early-career practitioners who build visibility, verifiable skills, and a discoverable presence during the current shortage window will be the ones holding the leverage when the gap begins to close. That window is open now.