thesurlydev

I'm curious how this is done effectively assuming no source should be sent to a model hosted remotely. Are there foundational model toggles for using inference and embedding but not using data for training?

I'm a software engineer and have embraced using AI to make me more productive. Seeing the fast pace of innovation and improvements of agents and tools leaves me wondering what I should shift my focus to.

Beyond the obvious answer of "learn how to use AI," I'm curious how other people would answer this question.

My own answer: Use AI tools to make the transition from employed software engineer to being able to produce value while transitioning to becoming self-employed.

I generally do a lot of reading online but I've slowly started to question my sources.

It's becoming increasingly difficult to find community without wading through the endless embedded ads and outright self-promotion or the culture of "whatever you're doing, it's not enough."

Thankfully, HN has remained pretty steadfast as a higher signal to noise community or at least that's my perception. What other communities do you frequent that have a higher signal to noise than most?

This has been bugging me for a long time. I have multiple Chrome profiles and a browser window open per profile. Inevitably, I click on a link in a different app and it attempts to open the link in the wrong browser window. The default seems to be the last window used.

Is there a solution which would allow me to open links in the "correct" browser window? This would include allowing me to define url patterns and map them to the correct browser profile (window).

I've tried countless tools to author OpenAPI specifications but haven't found one that I really like.

What's your favorite tool to author OpenAPI specifications and why?

I've recently joined a consultancy and a big part of the work is helping clients to level up extremely old apps. Other typical tasks include replacing old SOAP calls with newer HTTP/JSON microservice calls.

For a flavor of what I'm talking about a typical Java app is written in Java 6, is built via Ant and is manually deployed to machines in a private data center. There are no existing tests. Many of the library dependences are unversioned and the jars are checked into the version control system.

How would you approach the process of modernizing these apps?