AWS drew 120 participants to its DeepRacer competition in Auckland. Vehicular machine learning with 1/18th DeepRacer scale models. Photo / AWS
AWS drew 120 participants to its DeepRacer competition in Auckland. Vehicular machine learning with 1/18th DeepRacer scale models. Photo / AWS
OPINION:
Amazon Web Services, which needs no introduction, ran a DeepRacer event again in Auckland last week.
That involved a bunch of the cloud giant's commercial customers racing little radio-controlled cars with a camera mounted on them for image recognition, training the vehicles through machine learning (ML) to drive bythemselves on a track.
Congrats to the Auckland finals winners Sandfield and Vista, both of which go on to the ANZ regional finals held on Twitch. Three winners of the regionals are headed for Las Vegas, where they can compete for a US$5000 prize.
Not a lot of people know about DeepRacer, but it's a fun way to get started with ML of the reinforcement variety where you use rewards function to teach 1/18th scale cars that are powered by an Intel Atom processor and run Linux where to go.
It's easy to get started with DeepRacer in the cloud. You don't even need an Amazon account, just an email address for the login. The materials are slick, well-written, and easy to understand.
Even if you can't code, curiosity and patience (err, copy and paste then) should get you started until you've learnt more.
Now, AWS founder Jeff Bezos didn't become astronomically rich by giving stuff away for nothing, and the DeepRacer programme is more freemium than free.
There is a limited free tier with 5 gigabytes of persistent storage and 10 hours of ML modem training time, for 30 days. After that, it's US$3.50 per hour plus US$0.023 per GB and month.
Once you've cloud-trained and tried out virtual DeepRacer models, going physical with the 1/18th scale track car starts at US$399; the upgraded EVO car with light detection and ranging sensor and stereoscopic cameras for object detection is a shade under $1300 including shipping to NZ.
AWS drew 120 participants to its DeepRacer competition in Auckland. Photo / AWS
That's still affordable for companies, and DeepRacer is part of a bigger programme of learning materials, courses and scholarship funds that AWS has been running for a while to develop skills suited to its cloud.
Microsoft and Google have had similar initiatives too for years, like digital apprenticeships to expand the pool of new hires, and to upskill existing consultants and partner organisations. Overseas, there is Amazon's extended DeepRacer for students - and that looks set for New Zealand too.
Because of open-source software and interoperability demands from customers wanting to use multiple, competing clouds, the training initiatives are less self-serving than you'd think, with many of the skills picked up being transferable to other businesses.
For AWS which is set to launch a region in Auckland with multiple data centres and 200 staff to run them sometime in 2024 and more needed after that, educational efforts are hugely important for their business, and hiring efforts in a global, internet-scale market.
Furthermore, from a political perspective, incoming tech giants have to upskill locals.
Importing IT workers would be wildly unpopular to put it mildly, without training locals to do the same jobs.
Finding skilled labour ready to risk moving here is also hard when salaries are way higher just across the Tasman where globally mobile young people have more opportunities as well. That's before we start looking at the United States and Europe.
The flip side of the political coin is that inclusive training and hiring efforts onboarding Māori, Pasifika, women and gender-diverse people, who've long been subjected to discriminatory sink-or-swim policies, will be welcomed, if only because they make a difficult problem less visible.
Nothing stays the same forever though. After years of mad expansion, tech is taking a breather, as startups with insane valuations shed staff, and the giant crypto currency/blockchain/NFT pyramid scheme is finally falling apart.
That should alleviate and direct demand for workers towards tech sector areas that are anchored in reality. In other words, those that embrace and extend the economy rather than act as speculative parasites on top of it.
Market vagaries like the above apart, nothing short of a total financial meltdown worldwide is likely to dampen the need for continuous IT training. For anyone looking to change or to build a new career, it's worth keeping an eye on what the tech giants are offering.
At the same time, it's important not to write off traditional learning. Never forget that today's global tech companies were built on easy and affordable access to higher education, which doesn't, and shouldn't, necessarily align with commercial imperatives.
