HEADCOUNT & CODE · A Building Tech Teams publication
00 · HEADS UP - THIS IS A LONG ONE
I’ve started using WhisprFlow, and it seems significantly easier for me to rant my thoughts into a microphone than to type them out. So this newsletter is a little bit longer than the past ones. But, as always, these are three of the most interesting articles I’ve read in the past week, and my thoughts on Building Tech Teams.
01 · THE PATTERN
In the past two weeks, I’ve interviewed more senior-level technology roles than any other two-week period in the past two years. This year, I decided to get back on the tools and have more conversations with senior technology professionals. I was really driven by my interest in the changes AI is bringing - how it is changing roles and how companies are dealing with it.
My biggest reflection the past two weeks is that the difference between average and excellent is wider than ever. I’ve had people use AI on their CV to make themselves look better and more experienced than they were, and then a few minutes into the interview it comes out that basic-level Cursor or Claude Code is the extent of their AI engineering experience. On the other hand, I’ve talked to people who are running fleets of agents coding for them throughout the day, testing, building, reviews, evals, the whole shebang.
And what’s standing out - quality software engineering fundamentals and knowledge of quality software architecture are even more important now in the age of AI than they’ve ever been. Technical depth matters even more, and the three articles I’m sharing today show the pattern. Tech leaders are moving back closer to the tools. Tech leaders want to be closer to the product, understanding what’s being built and for whom, and the fundamentals of software engineering are proving to be one of the core differentiators for the AI engineers of the future.
I’ve learned this in the interviews. I’ve learned this even more from trying to build my own products using Claude Code, coming from a non-technical background. One thing is for certain: Vibe coding won’t get you the Senior AI Engineering roles.
02 · THE MOVES
1. Six in ten CTOs told Orosz they are on the way out.
Gergely Orosz interviewed close to twenty engineering leaders who are either on a career break or seriously considering one. Of the CTOs, VPs of Engineering, heads of engineering people he asked privately, six in ten said they are on the way out. He ranks ten reasons, and number one is that the job got much worse. (The Pragmatic Engineer, 19 August)
The Senior Engineering job is changing. Expectations are higher for technology managers and top technology professionals to know everything about AI, ship AI, and transform businesses. A lot of those expectations come from people who don’t actually know that AI is simply a tool, or where AI is going to help transform a business and what problems it is going to fix. Those expectations are being put onto the engineering leaders anyway.
Gergely’s top ten line up with my own experience this past week. I interviewed eleven technology leaders in a single day last Monday, and the same themes were consistent with Gergely’s findings:
One told me the role has lost its excitement. He isn’t actively looking; he likes the team and the culture, and he ruled himself out of a bigger role previously, on values.
Another has spent twenty years mostly in CTO roles at startups and scale-ups, describes himself as a professional problem solver who moves on once a thing is built, took a pay cut to join his current employer, but the equity stake isn’t going to be worth hanging around for.
A third is deliberately leaning technical over leadership, because he wants depth rather than a management ladder.
A fourth is motivated to leave by budget constraints, shifting priorities and red tape slowing delivery.
Gergely’s top ten also line up with the podcast discussions we’ve had on the BuildingTeachTeams podcast;
Number two on Gergely’s list is the startup losing and equity becoming worthless. That’s really interesting, because it lines up with the Matt McFarlane conversation on the last episode of our podcast with Matt talking about the commitment to actually see through a startup to an exit: “I saw some numbers from Carta the other day saying that the average duration it takes for a startup to exit or list or whatever has blown from eight years out to like 15 or something. So that horizon is huge now. That’s a third or a quarter of your career.”
Number eight is that AI individual contributors are being paid more than most traditional engineering executives. Adam Witanowski’s experience provides this, with his final few job offers being over half a million dollars a year as an IC.
Number nine is quitting to launch their own business. More and more technology professionals are considering quitting and launching their own business because AI makes it easier than ever to get a product to market. Cloë Stanbridge covered this on the podcast. AirTree run a programme called Frontier that writes founders a $250,000 SAFE cheque so they can leave their day job, and puts them in a cohort with a day a week in the AirTree office. When they opened applications, they got four or five hundred submissions. “Which is insane,” she said. “It’s given people the motivation and the ability to be able to do that where they just weren’t able to do that before, because it would take too long to build a product and test it and go to market.”
Underneath a lot of the conversations I have had this year is how to attract the best technology professionals, and it comes back to being a company that truly wants to drive change. The best technology professionals are driven by mission and vision, not just a role.
For leaders: if you are in a CEO or a board position, the point that stands out here is Orosz’s own: “Most of these companies’ EPD orgs will never go AI-native, not even close. Most VPEs aren’t good enough at change management to pull it off.” If you are serious about implementing AI in your organisation, you need to invest properly into it and into the people to execute on it. Unfortunately the commonality at the moment is “let’s do something with the AI”, and high expectations engineering teams, without having measured what we are trying to achieve.
For tech professionals: continuing to lean into the tooling and into becoming AI native gives you more opportunities. More opportunities to build for yourself, and to join startups, scale-ups and engineering teams who are genuinely changing the ways of work.
2. Canva changed how they build product, and who they hire.
Cliff Obrecht said most of the product development machinery that existed at Canva even a year ago has been abandoned. Teams once came with customer research, pitch decks, staffing plans and milestones. Now Obrecht wants “a prototype and the smallest team possible” before committing more resources. (Capital Brief, 19 August)
This is not new. This is the lean startup (one of my favourite books) movement made popular by Eric Ries.
But what’s interesting about Canva is that the philosophy is also changing how they hire and who they hire. Cliff quotes in interviews that candidates are asked, “How do you think about building products, and if it’s not that way, then you’re not coming in.”
It also speaks to the changing nature of product building as a whole. John Collison from Stripe shared with Cliff a similar effort at Stripe, to “deprogram” employees from over-engineered ways of building products.
This is as much a mindset change for product people and engineers as it is a tooling change. The tools have made the pathway to build, test and iterate faster and easier than ever. Builders need to be ok with version one not being perfect, but run lean and test before overcommitting.
Another interesting part of the conversation was the discussion of designers. We spoke about product people in last week’s newsletter. Obrecht argued that professional designers are increasingly becoming “brand architects” who set the rules and visual systems within which AI operates. In his words: “There is so much AI slop out there that you really need custodians of the brand in organisations.” This is the same move that happened with product people. Whilst product and design have been spoken about publicly as being replaced by tooling, two well-respected technology voices in two weeks have talked about the actual need for quality product and design people. Cagan in an article in Issue #8, and Obrecht here in a conversation.
For leaders: I think it is important to interview for the people you want in the door and for your ways of working. Make sure your interview reflects the way they will actually work in the job. Canva is doing it here in product and we spoke about Coinbase’s new engineering interview process in last week’s newsletter.
For tech professionals: be aware of the interview process going into a company, and what it may say about the ways of working, the type of people within the organisation, or the way they build / design / engineer. It is the best indicator of how the team might actually work, and whether that is suited to your way of working.
3. Fundamental skills for AI Engineering AND Knowledge workers.
Andrew Ng produced the AI Engineering Skills Map strictly in the software sense, naming four skills: building and deploying AI applications, software engineering fundamentals, using coding agents, and shaping the build.
From my experience, the fundamentals point is the most important. In Ng’s words: “even when you aren’t writing the code, understanding trade-offs between cost, scalability, reliability and speed produces better decisions on stack, architecture, data stores and testing.”
This has been my entire last two weeks. I have interviewed over 35 people who applied for senior AI engineering roles. Every great AI engineer I have interviewed has strong software engineering fundamentals. They understand how to architect solutions, where their knowledge is most valuable, and where to let the agents run.
The Daily Brief then extends the map to Knowledge Workers: five skills sitting on a foundation of domain judgment. AI capability mapping, context and harness management, problem and product prototyping, new opportunity identification, and rapid new skill acquisition. (The AI Daily Brief, 18 August)
This is the theme that has been running through the AI Jobs Index: the need for most tech jobs to incorporate AI skills, regardless of whether you are an engineer.
That last one, rapid new skill acquisition, is as important as ever, no matter your role. Professionals who keep growing their skills and learning new tools and ways of working will be the ones who stand out in their companies.
For leaders: For as much some things change, alot of things stay the same. Hiring for core fundamental skills as well as people who show a history of growing and learning sets you up well for now and into the future. Interview for what individuals are doing in their own time to progress their skills.
For tech professionals: I would continue to lean into new ways of working to become more efficient. I’ll give you an example of one of the best AI engineers I have spoken to in the past week. When it comes to setting the AI agents up to succeed before they start, he begins each day talking to his LLMs for two hours, providing context, questioning their decisions and defining a plan. He then lets them build for hours on end, and does a review session at the end of the day. It’s his context + tooling getting the job done.
03 · FROM THE INDEX
One read a week from the AI Jobs Index, the live tracker I run at aijobsindex.com.au.
The diffusion index reads 37.61 per cent. That is the share of technology and data-adjacent Australian job ads scanned that mention AI or an AI-specific skill, measured across the trailing thirty days. A fortnight ago it read 34.41.
This trend has held pretty consistently over the past few weeks: more technology jobs are asking for AI skills as part of what they’re looking for. And no new companies are laying off people because of AI.
Rather than continuing to profess the same numbers week in, week out, I’m in the process of securing new data sources to be able to accurately talk to:
Salary trends
Growth areas within technology and AI roles
Which states are hiring more
Other factors pertaining to AI’s influence on jobs in Australia
04 · WHAT I’M BUILDING
Last issue, I told you the AI Jobs Index was getting a team of agents for the rebuild, and that phase one was complete. Here’s what phase one actually did...
The job that pulls new listings into the Index had been dead for nine days. It broke on 16 August and I only fixed it yesterday.
The agents caught all of it - that’s the embarrassing part. The snapshot agent wrote “every figure above is carried over from last week, not re-measured” into its own report on Sunday night. A second one flagged the upstream error, and a third has been reporting, correctly, that a different job has been returning SUCCESS while writing zero records for more than two weeks. Every one of those messages arrived. Nothing happened, because I was in back-to-back interviews and the fix needed me.
So the thing I built to stop the Index serving zeros worked exactly as designed, and the Index served stale numbers anyway.
Nine days of drift was fixed with thirty-seven seconds of my attention and one command. But it required me - which was not the system I was hoping for.
This is the reason why I truly believe strong software engineering fundamentals and architecture are vital to the success of senior AI engineering talent. Right now, one of the first things I interview for is fundamentals, because solid engineers with great fundamentals don’t make the mistakes I do. I’m a non-technical builder learning on my own, trying to architect whilst working with Claude and checking the architecture against ChatGPT for validity, and I still have my gaps.
The fixes are in, though, and the build continues. Let’s just say it’s a work in progress.
05 · THIS WEEK ON THE PODCAST - MASHUP!!
I made this one for two reasons:
I think it’s really important to value the guests I’ve had on the show, and putting out their episode just once feels like I’m doing them a disservice, given the quality of the content that was in there.
And there was a really common theme around Salary that I wanted to make sure I could share.
This is the first of three mashups from the first eight guests. Nathan Hill from AWS, Claudia Barriga-Larriviere, Cloë Stanbridge from AirTree, Adam Witanowski and Matt McFarlane, cut from episodes 3 through 8, on what an AI-augmented technology professional is actually worth.
What we get into:
The question nobody can answer. I put it to Nathan Hill in episode 3: “I used to be called a principal software engineer. I was paid $200,000 a year. I’m now managing a team of agents. My productivity is 10x. What should I be paid?” His first response was “Aren’t you in the recruitment business?” Then: we can’t price it until we can measure it.
What the market actually pays. Adam Witanowski finished a nine-company process with seven offers. Senior AI roles here pay around $220k. The same work in the US was landing at $450k to $500k converted. What an Australian company needs to pay to be in that conversation: three to five hundred, “and that might feel ridiculous for a lot of companies”.
Whether the productivity is real. Claudia Barriga-Larriviere, on her own clients: “are they more productive, or are they just faster? I just don’t know if it’s the same thing.”
When the big cheque is worth it. Cloë Stanbridge has watched both versions run inside AirTree’s portfolio. Her line: “don’t hire someone who’s very expensive just because that’s what they were getting paid in Big Tech.”
Listen: YouTube · Apple · Spotify
Have a great week!
Cheers, James
Headcount & Code is the weekly read from Building Tech Teams. Written by James MacDonald, MD of NTP Talent. The AI Jobs Index is a Building Tech Teams publication.