Sabitlenmiş Tweet
Per Borgen
3.2K posts

Per Borgen
@perborgen
CEO of https://t.co/AF9LjT2054 (YC S20). Helping devs learn faster and grow their careers.
Oslo Katılım Nisan 2009
1.1K Takip Edilen20.6K Takipçiler


Throwback to when Reid Hoffman endorsed Jeffrey Epstein’s “skills” on LinkedIn
#ThrowbackThursday

English

@JoshuaEbner Roughly 60% of the open roles are located in the US according to Trueup, with SF being the biggest contributor.
SF -> 31k+ open roles
Austin -> 5k+ open roles
About a quarter of all jobs are remote.
English

@perborgen Any sense of where these rolls are located primarily?
I saw something by Jason Lemkin from a few months ago that noted that San Francisco generated something like 11,000 new roles in a recent one year period versus very few in smaller tech areas like Austin.
English
Per Borgen retweetledi

@PranteP27817 Indeed. That graph has been trending the right direction for two years now, even if the pace is modest.
English

@davidcasem @jack @stripe Same here, we’re busier than ever.
While AI speeds us up, we’re nowhere near running out of ideas for improving our products. If anything, we’re getting more of them.
English

Yeah, but AI is coming for your jobs. Honestly, shame on @jack for trying to deflect the blame for the 40% cut at Block to AI. Expect total chaos there. Use @stripe.
We've leaned heavily into AI. Our people are more capable than ever before -- and busier than ever.
We're hiring aggressively to meet our customers' needs.
English

Here’s something interesting I learned while working with AI APIs.
If you want a shorter response, a max_tokens limit might chop your text mid-sentence.
So you can give a simple instruction in plain language:
“Keep it under 100 words.“
But this isn’t a hard fix because LLMs aren't always great at estimating their own word counts.
Some models will only obey the limit if they have a reasoning step.
So you might get 97 words, or 124, or even a thousand.
Yes, a thousand. LLMs can randomly output massive responses.
If you don’t have max_tokens set, you’ll pay for every extra token generated.
The wise choice is to tweak both params and instructions if you’ve got response length requirements.

English

"Some super devs use the boost in productivity to slack off because they still produce more than their peers"
This is one of the biggest unanswered question wrt the AI productivity boost IMO, and not just for devs, but for all knowledge work:
How much of it will just be eaten up by more doom scrolling?
I suspect quite a lot.
English

Some factors:
* Most developers are getting started with AI
* Extremely few are running fully or semi automated pipelines
* There will still be demand for converting legacy software to AI enabled software
* Some super devs use the boost in productivity to slack off because they still produce more than their peers
English
Per Borgen retweetledi

@Joao_flashy "Never let the truth get in the way of a good story"
English

@perborgen Please don’t bring real data, people don’t like that.
Developers are dead. SaaS is dead. Everything is dead.
English

@aliwhosane It's still massively down since 2022 when money was free. But it seems that it bottomed out in 2025 for software developers.
Not for other jobs unfortunately.
English

@perborgen What is the rate of growth per multi year.. that would be a more accurate measure
P.S. - I don't believe devs are going anywhere anytime soon
English

@DevPriti1 Yes, these are real job postings from the largest job board website in the US.
While jobs postings have been decreasing steadily since 2022 when the ZIRP era ended, the trajectory for software developers has shifted over the last year.
English






