State of the Market

[gables]

London banking is dead. They don't seem to be hiring anything including IT. This has been the case for 3yrs.

When the banks start hiring, then everyone will start hiring. Clearly none of them have invested much into AI, knowing it will be just a other boom and bust cycle.

Not sure about dead for 3 years as I've been working in banking since 2022 which was my first time (this century) bar some time off last year on my part, but as I say below it helps if you have contacts.

As to AI, it appears here it is being invested in and heavily.

Its all outsourced nowadays all the big indian firms have pretty much all of the banking business in both the square mile & canary wharf. Last year the banks made tens of thousand of perm IT, back & front office workers redundant in canary wharf alone.

Saves them a lot of money in pensions, holiday pay & lower overall staffing costs....short sighted flawed strategy though as at some time they will need to start hiring experienced perm staff again...

Indeed... many people, permie and contractor, were let go and I was one of them last June, now back having started in mid-March - I think a lot is about who you know.

[uk contractor]

I have mentioned this before but I worked at a large European bank and their strategy in Canary Wharf at least was clearly based on cheap Indian developers. The dev floor I worked on was at least 70% Indian and half used to disappear at 4:30 in the summer to play cricket.

Yep sounds exactly like where I worked they had loads of poorly paid IT staff from the big indian outsourcers...many got paid £25-30K LOL yet they were being charged out at £500 ish a day.....replacing perms on £75-80K plus benefits + bonuses!

Most of the london based ones had to constantly call the mothership based in india as they were very inexperienced & more like resource to sit on the seat than actual proper IT staff....but line management at the bank did not care as they were under extreme pressure to cut headcount budgets or be part of the next wave of redundancy wave!

Any decent on the ball government would cancel all their visas force employers to hire locally born workers like they do in other countries when times are tough....

[uk contractor]

Not sure about dead for 3 years as I've been working in banking since 2022 which was my first time (this century) bar some time off last year on my part, but as I say below it helps if you have contacts.

As to AI, it appears here it is being invested in and heavily.



Indeed... many people, permie and contractor, were let go and I was one of them last June, now back having started in mid-March - I think a lot is about who you know.

AI is a dead end but it will take a few years maybe a decade before its obvious to the mainstream.....its best use is automate tasks the learning (LM) side is where all the hype sits though....LM never ever going to replace the human brain that's just the hype machine working overtime!

Where I was they axed nearly all experienced perm IT staff early last year...kept some lowly paid contractors & the brown tongue brigade! I have not bothered going back to the industry despite having a number of contract only offers I am not interested in doing a few months here & there as it will not lead to more perm employment as the industry trends are only moving in 1 direction right now.

Hedge funds & private wealth companies are still retaining perm IT staff though but have fewer openings.

[gables]

AI is a dead end but it will take a few years maybe a decade before its obvious to the mainstream.....its best use is automate tasks the learning (LM) side is where all the hype sits though....LM never ever going to replace the human brain that's just the hype machine working overtime!

Where I was they axed nearly all experienced perm IT staff early last year...kept some lowly paid contractors & the brown tongue brigade! I have not bothered going back to the industry despite having a number of contract only offers I am not interested in doing a few months here & there as it will not lead to more perm employment as the industry trends are only moving in 1 direction right now.

Hedge funds & private wealth companies are still retaining perm IT staff though but have fewer openings.

I am sceptical about AI and can certainly recognise the hype, I was merely pointing out that there seems an investment in it if that floats your boat, not for me though.

IMHO this type of hype has been a plague within IT all my career since 1989, and although I may have ridden the crest of some of it over the years I don't engage in the hype as such.

[willendure]

AI is a dead end but it will take a few years maybe a decade before its obvious to the mainstream.....its best use is automate tasks the learning (LM) side is where all the hype sits though....LM never ever going to replace the human brain that's just the hype machine working overtime!

Lol. The pace it is advancing at, give it a decade and it will be rewriting itself, at the double exponential improvement rate described by Ray Kurzweil in The Singularity is Near.

I don't know if you know about the new analog AI chip made by a start up called Extropic ? It is thought that a digital computer with roughly the same processing power as a human brain would need a 10MW data center using todays digital tech. Because multiplying numbers digitally is a very inneficient way to model nature. The Extropic chip shapes transistor noise into a probability distribution that it can sample way more efficiently. First chip they made was on aluminium and needed to be cooled down to low kelvins, the latest one is silicon at room temp. This is still some way off mainstream but it shows where its all headed.

If we ever get quantum computers to work, the tensor algebra describing entangled quantum systems is almost the same as that describing a neural net. I find that idea very scary as it would be orders of magnitude more energy efficient than us! I think maybe quantum computers are going to prove harder to crack than some might think - but the intermediate noisy AI stuff might help us get there once we get onto the self-improving path.

I have written about 200K lines of code with AI in the last 5 months - very well tested too I should add in case you think it might just be vibe coded tulipe. About 14000 test cases currently. I made an estimate of how long this would have taken me to develop by hand and its something like 2.5 to 3.5 years worth of work! Its an optimizing compiler complete with runtime and custom garbage collector.

I don't think AI will put software engineers out of work. Its a productivity multipler but one that everyone has. So everyone will need to use it to remain competitive, and will still need a head count to drive it. All that will happen is that we will be able to tackle larger and more ambitious projects. A bit like when the industry moved from assembly language to compiled languages - there was a boost to software size and complexity, but there was also a demand for people with the new skills.

As a contractor the rational way to respond is to upskill and remain competitive. Or find a niche that will never die - before King Billy reminds us how sweet the water is in Z-systems world.

[TheDude]

Lol. The pace it is advancing at, give it a decade and it will be rewriting itself, at the double exponential improvement rate described by Ray Kurzweil in The Singularity is Near.

I don't know if you know about the new analog AI chip made by a start up called Extropic ? It is thought that a digital computer with roughly the same processing power as a human brain would need a 10MW data center using todays digital tech. Because multiplying numbers digitally is a very inneficient way to model nature. The Extropic chip shapes transistor noise into a probability distribution that it can sample way more efficiently. First chip they made was on aluminium and needed to be cooled down to low kelvins, the latest one is silicon at room temp. This is still some way off mainstream but it shows where its all headed.

If we ever get quantum computers to work, the tensor algebra describing entangled quantum systems is almost the same as that describing a neural net. I find that idea very scary as it would be orders of magnitude more energy efficient than us! I think maybe quantum computers are going to prove harder to crack than some might think - but the intermediate noisy AI stuff might help us get there once we get onto the self-improving path.

I have written about 200K lines of code with AI in the last 5 months - very well tested too I should add in case you think it might just be vibe coded tulipe. About 14000 test cases currently. I made an estimate of how long this would have taken me to develop by hand and its something like 2.5 to 3.5 years worth of work! Its an optimizing compiler complete with runtime and custom garbage collector.

I don't think AI will put software engineers out of work. Its a productivity multipler but one that everyone has. So everyone will need to use it to remain competitive, and will still need a head count to drive it. All that will happen is that we will be able to tackle larger and more ambitious projects. A bit like when the industry moved from assembly language to compiled languages - there was a boost to software size and complexity, but there was also a demand for people with the new skills.

As a contractor the rational way to respond is to upskill and remain competitive. Or find a niche that will never die - before King Billy reminds us how sweet the water is in Z-systems world.

Where your test cases AI generated? If so maybe your code isn't as well tested as you think it is...

[willendure]

Where your test cases AI generated? If so maybe your code isn't as well tested as you think it is...

To some extent. I document invariants expicitly and derived test logic from those. The invariants themselves are very carefully hand curated but it is entirely possible that AI can write a bad test for them - I know in fact that it has sometimes done so. But you tend to find out when you see a bug and ask what invariant does this bug break ? Oh hang on, why did the test not catch it ? because AI only wrote half the test then gave up !

My claim of 14000 test cases actually masks what is really going on. My tests are split into test cases (code to be compiled) and test logic (copiler invariant checks) and every test case is run through every test logic. It is grouped into 14000 test cases in the test runner, but in reality its something like 150000 individual checks being run when you multiply it out.

The test cases are just code, which when compiled must always obeye all invariants that apply to it. Doesn't matter what it is - even code with type errors obeyes invariants up until it fails to type check and so on. Those are entirely AI generated which an amazing time saver. I can create a test coverage report, ask AI to look at at and try to dervive new test cases that will hit corner cases not already tested so leaving holes in the coverage.

[sadkingbilly]

how sweet the water is in Z-systems world

[Dion13]

I decided to accept a contract with 45% lower day rate (mush less senior role) than what I had in the last 5 years, just to avoid staying idle...

[ladymuck]

I decided to accept a contract with 45% lower day rate (mush less senior role) than what I had in the last 5 years, just to avoid staying idle...

Even if you have a healthy war chest, there's no point burning it if you can get some money coming in.