moonarchia
The Scientific Shitlord
- 29,817
- 57,970
IT/Software career thread: Invert binary trees for dollars.
- Thread starter Traak
- Start date
Kithani
Vyemm Raider
- 2,054
- 2,875
TJT
Mr. Poopybutthole
- 46,543
- 127,167
Welp I succeeded at getting a job offer... for a job I don't even want. It would legitimately be going back 10 years in terms of tech stack and I would be helping them convince the rest of the org to modernize.
Meaning I would barely do shit other than be bored making POCs and shit. Would be a good raise too. I just can't see myself doing.
It does sting to walk away from a $200k base, 15%/yr bonus and even a $20k sign on bonus that I didn't expect. But I just can't. But seriously their major problems are things I engineered out of my current job years ago and don't even think about. They're just beginning to think "hey maybe we need that this sucks."
Meaning I would barely do shit other than be bored making POCs and shit. Would be a good raise too. I just can't see myself doing.
It does sting to walk away from a $200k base, 15%/yr bonus and even a $20k sign on bonus that I didn't expect. But I just can't. But seriously their major problems are things I engineered out of my current job years ago and don't even think about. They're just beginning to think "hey maybe we need that this sucks."
Khane
Got something right about marriage
- 21,540
- 15,443
I can understand all that for sure. There is a bigger behemoth lurking in the depths of tech for us though.
5 years from now how much damage will AI "vibe coding" have done to the majority of companies and what kind of mess would you be walking into elsewhere? This company at least seems to be resistant to that sort of mess from what you're saying.
The company that I currently work for, have worked for since 2007, is deteriorating at such a break neck pace because of this that I wouldn't be surprised if it was sold off for parts by our parent company within the next 5 years.
- 1
Noodleface
A Mod Real Quick
- 39,432
- 17,917
We got a developers survey today which just went to show how out of touch our company is with what I do. One of the questions was about how long it takes to go turn around a code change and test it, options were something like 10s, 30s, 60s, 2 min, 5 min and that was it.
Our build takes more than 5 minutes... A server might take 20-30 minutes to flash our image. Testing is not immediate either, we may need to use hardware emulators/debugging tools.
AI is being sold to companies and I don't think a single exec is grounded in reality.
Our build takes more than 5 minutes... A server might take 20-30 minutes to flash our image. Testing is not immediate either, we may need to use hardware emulators/debugging tools.
AI is being sold to companies and I don't think a single exec is grounded in reality.
- 1
- 1
TJT
Mr. Poopybutthole
- 46,543
- 127,167
Extremely unlikely. Mostly I'd be helping sell the greater organization on this or that and squeezing it in where I can, if at all. They had very specific interview questions about that, "how would you sell this to a large team of stakeholders?" Along with a small sample pitch of something and other nonsense. For a staff engineer role.
- 1
TJT
Mr. Poopybutthole
- 46,543
- 127,167
I think beyond that it fundamentally ignores that the means to teach AI (human documentation) doesn't read in a truly step by step machine fashion. A manual for hardware specifics, database rules, and so on is written with the implicit understanding that the person reading it has some background in it and would fill in the gaps. AI cannot do this which leads to bizarre behavior in general.
Like what you touched on earlier about the AI not understanding the boot sequence and thus coding incorrectly. Even though it can indeed spit out 500 lines of code in an instant. Similarly in the data space, unless you spell out how relationships between objects and databases are defined it just does nonsense. Made more complicated by most modern database systems not even having declared relationships like SQL Server does, and even SQL Server doesnt have all of that accessible to the AI to work with. So you're left babysitting the AI to get the answer you want and you maybe saved 10% of your time doing it.
- 1
moonarchia
The Scientific Shitlord
- 29,817
- 57,970
TJT
Mr. Poopybutthole
- 46,543
- 127,167
Yeah they are currently ignoring the freemium AI prices.
AI are saas products. Everyone gets the teaser rates right now but in another year or two these companies expect to close the trap. Would you look at that your entire staff can't even do day to day tasks without AI-Slopmaster 7.2 on Enterprise ChatGPT? Then they squeeze you. Just like Salesforce, ServiceNow, AWS, and so on all do.
That 10% of time is about to get real expensive. But its like 8 quarters out so leadership doesn't even realize it exists.
moonarchia
The Scientific Shitlord
- 29,817
- 57,970
Sheriff Cad
scientia potentia est
- 31,181
- 74,047
Assuming this doesn't go long enough at Freemium prices (great term btw) for people to retrain into other things, doesn't this just pop back to equilibrium when the prices become the "real" prices? When companies have to pay more for AI than they paid for the people, and the AI is producing less, wouldn't they just rehire people and find some balance?
TJT
Mr. Poopybutthole
- 46,543
- 127,167
Not with the current pricing models for... everything. Microsoft are the Grandmaster dickheads of charging per seat, per instance.
Now you have Excel Jockeys that can't Excel without also having Claude Code Excel plugins that also cost $$ ontop of that. So where do you save money other than trying to pay the humans less?
Control
Golden Baronet of the Realm
- 5,477
- 15,583
What do you make of china dropping open-source models that are approaching frontier models, at least in benchmarks. Running them locally is out of reach of consumers of course, but a slightly worse model, a beefy ai server, and someone to keep it from falling over (not to mention actual control over your data) may start to make a lot of sense when prices get jacked up.
Sheriff Cad
scientia potentia est
- 31,181
- 74,047
But those are the trends because thats currently cheaper, no? If running on their own metal was cheaper, they'd be doing that, wouldn't they?Well then you're back to whatTomServo just said. You need metal to run all that shit. Which firms are apparently dumping. So they're going to run it on AWS I guess and get spitroasted?
TJT
Mr. Poopybutthole
- 46,543
- 127,167
I haven't yet heard of any company opting to run open source AI on their own datacenter or AWS to save money paying Claude Code and the like.
Today, right now, paying the freemium rate is LOTS cheaper. But that is a fuse on a bomb. One just long enough that leadership today doesn't see what's happening.
Today, right now, paying the freemium rate is LOTS cheaper. But that is a fuse on a bomb. One just long enough that leadership today doesn't see what's happening.
- 1