Every AI recruiter in Australia will tell you the market is short of talent. Almost none of them will tell you that the shortage is not the hard part of your hire. The hard part is judgement: knowing which of the several hundred people who now describe themselves as machine learning engineers can actually put a model into production, and knowing what that person will accept in a year when the cash rate sits at 4.35 per cent and general wage growth has drifted down to 3.2 per cent. Sourcing is the cheap part of AI recruitment. Filtering is the expensive part. If you are asking what you should know before engaging an AI recruiter anywhere in Australia, that is the whole answer in three sentences, and the rest of this piece is the evidence.
A disclosure first. We are Big Wave Digital, a Sydney recruitment agency founded in 2010 by Keiran Hathorn. We have been placing AI talent since 2021 and we compete with several of the firms mentioned below, so read what follows with that in mind. We are Sydney-based; most of our AI work is in Sydney and Melbourne, and we say so plainly because a national piece written by a Sydney firm should.
Is there really an AI talent shortage in Australia?
The broader labour market is cooling. The ABS Labour Force release for July 2026, published on 20 August, put the seasonally adjusted unemployment rate at 4.5 per cent, with employment falling by 16,000 people over the month and the participation rate slipping 0.2 points to 66.9 per cent. Underemployment sat at 6.4 per cent. This is not a market in which most employers struggle to attract applicants.
AI hiring lives inside that market but does not behave like it. What we see, and what any honest AI recruiter will confirm, is a glut of applicants and a scarcity of proven people. Since 2023, a large share of data analysts, a good number of software engineers and a surprising number of product managers have added the two magic letters to their titles. Post a senior machine learning role in Sydney or Melbourne today and you will receive hundreds of applications within a week. The fraction who have shipped a model that real users depend on is small, and it does not get larger because the pile does.
So the shortage is real, but it is a shortage of signal rather than of bodies. The useful analogy is a gold rush: there was never a shortage of dirt, only of people who could tell which pan was worth the effort. A recruiter who sells you access to the dirt is selling something you already have. The scarce skill is the panning.
Why does a cooling economy not cool AI salaries?
Here is the second thing nobody tells you. The ABS Wage Price Index for the June quarter 2026, released on 19 August, showed annual wage growth of 3.2 per cent across all sectors and 3.1 per cent in the private sector, with 79 per cent of jobs recording a wage change of less than 4 per cent over the year. On 11 August the Reserve Bank left the cash rate target at 4.35 per cent after three increases earlier in the year, and said plainly that it does not expect inflation to return to around the midpoint of its target until late 2027. The same statement noted that growth among Australia’s major trading partners has been propped up by AI-related investment.
Put those two facts together and you get a two-speed market. Most Australian wages are settling. AI wages are not, because the investment that drives them is global and the Australian supply of production-grade people has not caught up. Our own market ranges, informed by ABS earnings data and our placement experience, have held through 2026: mid-level AI and machine learning engineers at $130,000 to $165,000 plus super, senior engineers at $165,000 to $210,000, lead and principal roles from $210,000, and contract day rates of $950 to $1,400.
The consequence for your budget is uncomfortable. Finance teams are reading the same Wage Price Index and asking why an AI hire should cost more than a 3 per cent uplift on last year’s number. The answer is that the candidate is not being priced against last year’s number; they are being priced against three other offers. And a restrictive cash rate makes good candidates cautious about moving at all, which means counter-offers stick harder and persuasion takes longer. A wasted AI hire in this environment costs more than it did two years ago, not less, because the money is scrutinised more closely and the replacement is no easier to find.
The forty CVs nobody read
A founder of a logistics scale-up in Brisbane rang us earlier this year. The details here are changed, but the shape of it is exact. He had engaged a large national network to find a senior machine learning engineer. Within a week he had forty CVs. He was impressed; the CVs were polished and the recruiter was responsive. He hired the most articulate candidate, a person with a strong academic record and a portfolio of fine-tuned models.
Six months later nothing was in production. The engineer could fine-tune beautifully in a notebook but had never deployed a model behind an API, never dealt with drift, never been on call when a prediction service fell over at two in the morning. When the founder went back through the process, he realised nobody had ever asked the candidate three questions: what have you shipped, who used it, and what broke. The recruiter had not asked because volume was the product being sold, and volume does not require those questions. The founder had not asked because he assumed forty CVs meant forty screened people.
Oscar Wilde has Algernon say in The Importance of Being Earnest that “The truth is rarely pure and never simple.” A CV is the pure and simple version of a career. The truth of whether someone can build AI that survives contact with users lives in the messy parts, and the messy parts are what a genuine AI recruiter spends their time excavating. It is the work we did when we placed the first 20 AI team members at Leonardo.ai before its acquisition by Canva, and it is why the interesting question about any AI recruiter is not how many candidates they can find but how many they can rule out, and on what grounds.
How do you tell a genuine AI recruiter from a relabelled tech recruiter?
Australia has a growing number of firms that recruit in this space, and some do it well. Talenza, Kaliba, Brightbox Consulting and Talent International all place AI and data people, and the national networks such as Hays and Robert Half have reach that a boutique cannot match. The question is not whether a firm is large or small. It is whether the consultant in front of you can do the filtering, because that is what you are paying for.
Four questions settle it quickly. Ask them to describe their last three AI placements in detail: the problem the client had, what the candidate had built before, and what happened in the first six months. Vagueness here is disqualifying. Ask how they verify production experience, and listen for whether they interrogate deployment, monitoring and failure rather than tools and courses. Ask what the candidate’s salary expectation is relative to the market and make them defend it with something more than “that is what people are asking.” And ask what their offer drop-out rate looks like for AI roles, because a recruiter who does not track it is not learning from it.
There is a trade-off to be honest about. A national network gives you reach, which matters if you are hiring in Perth or Adelaide where the pool is thin. A specialist gives you judgement, which matters everywhere. Work out which of the two you are actually short of before you sign terms, and do not let a firm sell you reach when what you lack is judgement. If you want a longer treatment of the mechanics, our guide on how to hire AI engineers in Sydney covers the process step by step, and our comparison of AI recruitment agencies in Sydney sets out how the local firms differ.
Three golden nuggets
Ask for the shipped list before the CV. Tell your recruiter that for every candidate you want a single paragraph, written by them, describing what the person has put into production, who used it and one thing that went wrong, before you see a resume. This works because it forces the recruiter to conduct the interview you would otherwise have to conduct, and it separates notebook engineers from production engineers in minutes rather than weeks. A recruiter who cannot write that paragraph has not spoken to the candidate properly.
Agree the band in writing, super inclusive, before the search starts. With the Wage Price Index at 3.2 per cent and AI salaries not following it down, the gap between what your finance team expects and what the market clears at is the single most common reason offers collapse. Put the band on paper with your CFO and your recruiter before a candidate exists, and specify whether the figures include superannuation, because a $180,000 offer that turns out to be $164,000 base loses candidates on the spot. The negotiation you have internally in week one saves the one you would otherwise lose externally in week six.
Run the reference on the model, not the person. When you check references, ask the referee what happened to the system after the candidate left. Did it keep running, did someone else have to rebuild it, did anyone understand it. Models drift and pipelines rot, and the clearest sign of a good AI engineer is what they left behind that still works without them. Referees will answer this question candidly because it is not the question they were briefed to expect.
What to do this week
Pull up the last AI job advertisement you ran and count the lines. How many describe what the person will ship, and how many list the tools they must already know. In our experience the ratio is usually inverted, and that inversion is the reason the pile of applicants is tall and the pile of usable ones is short. Rewrite the advertisement around outcomes, keep the tool list to what is non-negotiable, and give it to whichever recruiter you use with the four questions above. You will learn more about that recruiter from their response than from any pitch deck.
If you would rather have that conversation with people who have been doing this since 2021, with an 89 per cent repeat client rate over 16 years and placements at Apple, Universal Music and Spacetalk, talk to us. You can also read more about our AI recruitment in Sydney and how we approach the filtering.
