Reposted by Eugene Meidinger
Big Tech’s AI push has tanked demand for juniors, even computer science Stanford grads, who lack the experience to vet AI code.
It’s a stark shift for a sector historically obsessed with youth and pedigree. I don’t think this is sustainable, but it’s unclear if or when the pendulum will swing back
It’s a stark shift for a sector historically obsessed with youth and pedigree. I don’t think this is sustainable, but it’s unclear if or when the pendulum will swing back
They graduated from Stanford. Due to AI, they can’t find a job
When they were freshmen, ChatGPT hadn’t yet been released upon the world. Today, AI can code better than most humans.
www.adn.com
December 29, 2025 at 6:39 PM
Big Tech’s AI push has tanked demand for juniors, even computer science Stanford grads, who lack the experience to vet AI code.
It’s a stark shift for a sector historically obsessed with youth and pedigree. I don’t think this is sustainable, but it’s unclear if or when the pendulum will swing back
It’s a stark shift for a sector historically obsessed with youth and pedigree. I don’t think this is sustainable, but it’s unclear if or when the pendulum will swing back
Working on a fun Claude session replay for an upcoming blog post
December 29, 2025 at 2:24 PM
Working on a fun Claude session replay for an upcoming blog post
Apparently Gemini CLI doesn't understand the concept of OneDrive and felt the need to try to index everything which caused it to try to download everything.
December 28, 2025 at 2:40 PM
Apparently Gemini CLI doesn't understand the concept of OneDrive and felt the need to try to index everything which caused it to try to download everything.
If I didn't have health insurance, I wouldn't be able to afford being on a pump.
December 28, 2025 at 12:51 PM
If I didn't have health insurance, I wouldn't be able to afford being on a pump.
Reposted by Eugene Meidinger
⚠️ I'm decommissioning dataplatform.social on March 31. ⚠️
Migrating your handle to another server (and bringing all your followers/followed handles) is relatively simple, but you have to do it while the server is still up.
Information & migration instructions:
dataplatform.social/@dhmacher/11...
Migrating your handle to another server (and bringing all your followers/followed handles) is relatively simple, but you have to do it while the server is still up.
Information & migration instructions:
dataplatform.social/@dhmacher/11...
Daniel Hutmacher (@dhmacher@dataplatform.social)
SHUTDOWN NOTICE: I'm planning on decommissioning the dataplatform.social server on March 31.
The server currently has less than 30 active users (with activity in the last month) and even though I've ...
dataplatform.social
December 27, 2025 at 10:59 AM
⚠️ I'm decommissioning dataplatform.social on March 31. ⚠️
Migrating your handle to another server (and bringing all your followers/followed handles) is relatively simple, but you have to do it while the server is still up.
Information & migration instructions:
dataplatform.social/@dhmacher/11...
Migrating your handle to another server (and bringing all your followers/followed handles) is relatively simple, but you have to do it while the server is still up.
Information & migration instructions:
dataplatform.social/@dhmacher/11...
Claude Code plus Skills plus Python can be a way to structure your thoughts in a methodical way and limit the amount of randomness the LLM introduces. I'm currently experimenting with a set of Claude Code skills to help me with course design from start to finish.
December 27, 2025 at 1:16 AM
Claude Code plus Skills plus Python can be a way to structure your thoughts in a methodical way and limit the amount of randomness the LLM introduces. I'm currently experimenting with a set of Claude Code skills to help me with course design from start to finish.
Visual DAX execution plan proof of concept, written entirely by Claude Opus over 3 days.
December 25, 2025 at 5:47 AM
Visual DAX execution plan proof of concept, written entirely by Claude Opus over 3 days.
Reposted by Eugene Meidinger
Join me and others for a remote conference in January.
I will demystify DAX in an 1h webinar and start you off in data modeling in a 4h workshop.
www.linkedin.com/events/m365c...
I will demystify DAX in an 1h webinar and start you off in data modeling in a 4h workshop.
www.linkedin.com/events/m365c...
LinkedIn Login, Sign in | LinkedIn
Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.
www.linkedin.com
December 23, 2025 at 9:13 PM
Join me and others for a remote conference in January.
I will demystify DAX in an 1h webinar and start you off in data modeling in a 4h workshop.
www.linkedin.com/events/m365c...
I will demystify DAX in an 1h webinar and start you off in data modeling in a 4h workshop.
www.linkedin.com/events/m365c...
Grimwald is sick of Copilot too, Bramble is a biiit niave.
December 23, 2025 at 3:54 PM
Grimwald is sick of Copilot too, Bramble is a biiit niave.
Shadowmere is also annoyed at all of these Preview features.
December 23, 2025 at 3:40 PM
Shadowmere is also annoyed at all of these Preview features.
Eugenebot asking the REAL questions.
December 23, 2025 at 1:50 PM
Eugenebot asking the REAL questions.
Using DSPy to find prompts at the pareto frontier of length versus accuracy is *fascinating*
In plain English: You can use an LLM to modify and test prompts and find the best ones for each tradeoff of result accuracy and prompt shortness.
In plain English: You can use an LLM to modify and test prompts and find the best ones for each tradeoff of result accuracy and prompt shortness.
December 23, 2025 at 12:36 PM
Using DSPy to find prompts at the pareto frontier of length versus accuracy is *fascinating*
In plain English: You can use an LLM to modify and test prompts and find the best ones for each tradeoff of result accuracy and prompt shortness.
In plain English: You can use an LLM to modify and test prompts and find the best ones for each tradeoff of result accuracy and prompt shortness.
If social media is going to be all bots, I might as well just build my own.
December 23, 2025 at 4:29 AM
If social media is going to be all bots, I might as well just build my own.
This is basically the core of politics
you can tell so much about people by whom they give the benefit of the doubt to
December 23, 2025 at 1:02 AM
This is basically the core of politics
I really thought I could make it through my career without learning Docker.
December 22, 2025 at 6:03 PM
I really thought I could make it through my career without learning Docker.
It's certainly not as *smart* as me, but I've successfully applied voice training to make Qwen *sound* like me.
December 21, 2025 at 3:34 PM
It's certainly not as *smart* as me, but I've successfully applied voice training to make Qwen *sound* like me.
This AI book by Welch Labs looks gorgeous. Such an amazing YouTube channel for learning.
www.welchlabs.com/resources
www.welchlabs.com/resources
Resources — Welch Labs
www.welchlabs.com
December 21, 2025 at 2:22 AM
This AI book by Welch Labs looks gorgeous. Such an amazing YouTube channel for learning.
www.welchlabs.com/resources
www.welchlabs.com/resources
Reposted by Eugene Meidinger
A good example of the kind of development AI tools are good at. A developer feeds an APK of an Android word game to OpenAI’s Codex + ChatGPT 5.2 and creates a working iOS version which is now in the App Store without writing a line of code.
He did get permission from the original game developer.
He did get permission from the original game developer.
Wordiest
I used to write more
blog.fsck.com
December 20, 2025 at 9:25 PM
A good example of the kind of development AI tools are good at. A developer feeds an APK of an Android word game to OpenAI’s Codex + ChatGPT 5.2 and creates a working iOS version which is now in the App Store without writing a line of code.
He did get permission from the original game developer.
He did get permission from the original game developer.
So @itsnotaboutthecell.com what if I used AI to cluster and categorize 3,800 comments so I can identify blog topics so I can stop answering the same question over and over?
Would that be a good idea?
Asking for a friend.
Would that be a good idea?
Asking for a friend.
December 20, 2025 at 8:01 PM
So @itsnotaboutthecell.com what if I used AI to cluster and categorize 3,800 comments so I can identify blog topics so I can stop answering the same question over and over?
Would that be a good idea?
Asking for a friend.
Would that be a good idea?
Asking for a friend.
Ground to Cloud SQL Server™ means I'm no longer safe from it
December 20, 2025 at 7:05 PM
Ground to Cloud SQL Server™ means I'm no longer safe from it
Sharing this classic because a colleague on LinkedIn is being a troll.
www.sqlgene.com/2018/08/12/d...
www.sqlgene.com/2018/08/12/d...
Doing the bare minimum to stop toxic behavior - SQLGene Training
I recently saw on Twitter some abusive behavior, and I know that Twitter and trolls go together like peanut butter and jelly, but for whatever reason I’m angry. Maybe because I know the person who was...
www.sqlgene.com
December 20, 2025 at 3:36 PM
Sharing this classic because a colleague on LinkedIn is being a troll.
www.sqlgene.com/2018/08/12/d...
www.sqlgene.com/2018/08/12/d...
Getting claude code to teach me how to use claude code
December 20, 2025 at 3:03 PM
Getting claude code to teach me how to use claude code
Well I can't ever make this Android repo public now.
December 19, 2025 at 12:53 PM
Well I can't ever make this Android repo public now.