Is it Time to Become an ML Engineer?

On the heels of the dot.com bubble, the term "labor arbitrage," appeared as a technical euphemism for what the general US labor force referred to with great anxiety as, "Outsourcing."  Thanks to changes in Clinton era policy and the emergence of the internet, it was easier than ever to employ abroad at lower rates.  In doing so, the hope was that this quid-pro-quo-ish arrangement would lead millions of latent foreign consumers to buy American products.  This would in turn lead to a global economic interdependence and subsequent great peace and prosperity for all.

Things didn't play out that way exactly.  China, India, Singapore, and others certainly did grow richer.  American businesses weren't given carte blanche to freely swoop in and recapture that wealth though.  In addition, arbitrage generally drives prices to equilibrium.  From the get-go, the outsourcing labor trade had an expiring lifetime.

Moreover, the trade goes both ways.  Ingenuity on the labor side meant talent migrated to where the money was.  If by moving to a generally well known and regarded country you could quadruple your salary, wouldn't you?  Conveniently, our government stepped in to introduce the H1-B, a new formal process for immigrant labor that aimed to "protect American workers."  It makes immigrant tech laborers beholden to their sponsor, reducing the liquidity of the labor market, allowing employers to pay less and demand more.  Wait...until this moment they were happy exporting American jobs but now they want to protect them?

Dismayed that high labor rates "persisted nevertheless," big tech tried other avenues.  Women in STEM attempted to rekindle the flame of feminism, acutely focused on STEM domain.  Since its inception, the feminism movement had roughly doubled the workforce, driving down wage inflation for employers.  Somehow it had failed to reach STEM though, where most employees where still male.

Why stop there?  DEI took this approach another step by extending the desire for equity beyond just women and to all underrepresented groups.  Universities using DEI as a metric for admission would ideally address inequity early on, well before employers' hiring decisions.  The intent sounded good.  Everyone wants to live in a more fair world of equal opportunity.  In forwarding this narrative, it certainly didn't hurt that Facebook, Twitter, and Bezos owned Washington Post were overtaking traditional mediums for news and pop culture.

Suddenly, with the second Trump inauguration, DEI is actively being reversed though.  With shocking lack of hesitation big tech dropped DEI.  Wait, what about social justice?  At a minimum, what about lowering labor costs?! What's going on here?!

In fact, before DEI got marched back, big tech had already receded their bet on human labor and placed their chips squarely on AI.  Once again, a panacea to the woes of paying high labor costs had arrived.  There's an ongoing doubling down that AI will replace STEM workers.  Big tech no longer has any need for DEI or H1B.

It'll be interesting to see how the big bet on AI plays out.  Like it or not, most of us have a large stake riding on it.  My bet is it'll wind up a lot like all of its predecessors, somewhere between either extreme.  Inching the world to a slightly more ideal place without quite panning out as the hail Mary payoff hoped for.

 The question for those of us working in Software Engineering is, "Is it time for me to make a lateral career move?"  An enormous amount of the free capital in the economy has poured into AI. Simultaneously, the Fed has raised rates with the direct intent of encouraging layoffs.  A combination of pressures are mounting against tech labor.
Just five years ago, parents were sending their kids to coding camps.  Now the coders themselves are left worrying about their own prospects.  The sanguine spirit of the 2010s has transformed into uncertainty. 

Capital is far more liquid than knowledge, people, or most assets.  Consequently, the US economy has a long standing tendency of rushing for gold.  In my three decades working so far, I've experienced it several times.  Some gold rushes turn into viable mid to long term careers.  Some wash away within a few years as passing fads.  During the Cold War aerospace engineering was in fashion.  The fall of the Berlin Wall and Dot.Com largely replaced this with a new wave of Software Engineering.  Its bust shook out many, only to be rekindled by "Software Eats the World."  That's now experiencing a similar capitulation to AI.

For a new technology the chronological progression to maturity seems to be hardware -> infrastructure/tooling -> product.  The capital investment really doesn't start paying off en masse until the Product phase.  Right now (Q1 '25), I'd say we're between the hardware & infrastructure/tooling phase as evidenced by the Nvidia trade.  Will capital be able to hold its conviction until the Product phase?  It depends on how long it takes to get there.  For the web, it was over a decade from Dot.com (~2000) to Software Eats the World (~2015).

 

 

My guess is we'll enter the Product phase between 2030-2035.  That may or may not be enough time for one more boom and bust cycle.  If you jump headlong into ML, there's a good likelihood of layoffs in the next several years depending on how the phase of the macro boom-bust aligns with the "trough of disillusionment" in the hype cycle.  Maybe it's worth the risk to accumulate the expertise.

As an engineer, you're making a bet either way.  On one hand, if you stick with a pre-AI focus and it shakes out, you may be left unemployable.  If however you hold on long enough, you may find that you're one of a small group that retained that skill and can charge a premium.  For example, there's a small center of excellence for mainframe programmers in the Philippines.  The global financial industry is highly dependent on them for maintenance of their securities clearing processes.

There's a reason "fortune" means both luck and money.  Personally, I'm focused on continuous learning of long term knowledge.  A good engineer is fundamentally a problem solver.  The economy will always need people willing and able to solve its hard problems.  If you're getting into engineering now though, I'd encourage you to be aware of the need to constantly re-skill and face these booms and bust cycles.  It can be exhausting.  If you don't enjoy the work it may be hard to weather long term.

https://frontierai.substack.com/p/you-cant-build-a-moat-with-ai-redux

https://toddle.dev/blog/why-is-everyone-trying-to-replace-software-engineers

https://blog.gregbrockman.com/its-time-to-become-an-ml-engineer

https://www.kalzumeus.com/2012/01/23/salary-negotiation/