Corewood

DataDog is like, "hold my water bowl..."

Corewood has new website content including the Machine Learning API services we are able to offer.

If you want to introduce #AI into your #API as efficiently as possible, with solid ops built in, you want Corewood.

corewood.io/services

We stumbled into making an #AI privacy #Platform.

Ooops. We wanted to be a #DevOps consultancy. We would have written your #AWS #infra and #Terraform modules.

But NO.... we got bored.

And then what happened? Well we got ok performance #ML inference @ Edge.

Now we're using it for #DataPrivacy.

We were #DevOps and #backend consulting. Then we were #HighPerformanceMachineLearning.

We need a niche, though, and we have chosen the niche that aligns with our opportunities and our values: #DataPrivacy

#OnPrem. #PrivateByDesign. #Secure. @corewood.cloud data privacy systems.

corewood.io

Proof for the skeptics.

It's been a crazy 3 months but we finally found our niche. We are a #DataPrivacy company. We are #ML #infrastructure and #LLM capable.

Convert #OpenSource models into #BusinessValue with Corewood.

</advertisement>

For real we got this working in #GoLang and it's super cool. ML-driven, no regex.

Don't know who to credit but this is gold 🤣

Me watching Cursor in Agent+ YOLO mode:
"No, it's not that, you dumb mach... oh, oh you got ther... oh that's a good way to... ok fine. Wait slow down I didn't catch all of that."

If you're using cursor, please run this command:
"Remove all private data from console outputs except for tests"

Pass it on.
#StandWithUkraine

Elebase is a geospatial data management platform with a powerful API. | Elebase Elebase is API-first content and data management for digital maps.

Want to hyper-charge getting your custom map application off the ground? Or do you ever want to use the same content you're putting in the map on your website or blog as well?

Wouldn't it be nice if you had a headless CMS to manage user data?

elebase.io

Put your content on the map.

#WeFixAICode

Remember, AI is a party trick. You know how a magician shuffles up a bunch of cards and does cool things that make the room go, WOW!!

But, there's still no such thing as magic. Sorry. The laws of physics and math make the machine "work."

There will be no "migrating" AI-generated code to different systems.

As with crufty monoliths, eventually they're so valuable AND so fragile that you have to grin, bear, and maintain.

If AI advancements teach us one thing: The more things change, the more they stay the same.

#WeFixAICode

#Contractors, #Consultants, #SMB...

What tips do you have for getting contracts / finding people who need the help you're able to provide?

#APIs the #OpenAPISpec are amazing; they enable you to monetize your business model in more channels.

But the real challenge of APIs is semantic. "How your APIs see your business" vs "How your customers see your business."
Can't document your way out.

Understanding semantic complexity takes time.

Everyone wants to do things better. Everyone thinks how they learned to do things is the right way.

The only test that works: outcomes.

Does it feel like there should be some code or etiquette around using #AI? Like, hey I used AI to write this.

Anyone else feel a bit deceived by AI generated sensationalist content on socials?

This is where I got to as well. Implemented x509 device identity -> JWT exchange, and the full-scale cert authority model was too heavy. Implemented a just-enough version that treated the cert as a cred and that worked smoothly.

If your business workflow requires code edits, it will not #scale.

"Maybe #terraform isn't the right answer to everything..."
The Industry: "#DevOps is OUT, #PlatformEngineering is IN"

..."So how do I modify the configuration?""
Engineers: "So you edit this file and `git commit ...`"
#facepalm

🔧 #DevOps truth: It's about building resilient systems, not permanent positions.

Most companies need DevOps knowledge periodically, not constantly.

Hire consultants for the expertise. Build automation for the day-to-day. Your systems should mostly run themselves.

#TechStrategy #Engineering

How did you learn computers?

You used to need to know more. Configuring drivers, backing stuff up to set up a new machine, upgrading Windows always took fixing.

Today, it seems like we have so many services and plug-&-play set ups that people don't spend their time fixing & cursing. I feel old.

Want to learn how to explain #DevOps, #NonFunctionalRequirements, and #SystemsArchitecture to friends, family members, and attractive people at cocktail parties? And sound compelling and interesting while you're at it?

Follow this link:

It's like getting interrupted mid-sentence with something you weren't going to say.

Would you say you're gonna... "kick the bucket?"

(For non-English natives, kick the bucket is an expression for dying. "Kick" is also an expression for giving something a try. It's a bad pun.)

While the actual cluster set up is simple, you have to understand the AWS ecosystem too.

All in all I find AWS and K8s to share most domain complexity.

There's only so much abstraction before you get into endemic complexity.

We use Podman over Docker. We didn't like the way Docker got about licenses and fees.

podman.io

That's a great point.

What makes you say it's a shame?

A tech person and business owner can appreciate how cheap self-managed k8s can be (done well). An ops employee for a big company does not want to get woken up to fix stuff, so public cloud.

I think there's a place for both approaches.

What did you find challenging?

I learned all the AWS things before I learned #K8s. I fell into the "it's complex" trap and put off learning K8s for years. When I finally worked full time on a K8s platform, it's no less complex than any system that's:
- secure
- load balanced
- autoscaled
- orchestrated
- extensible
- observable..

#DevOps
"Development Operations." Shouldn't that mean "how we operate software development"?

- Builds & deployment
- Automated testing
- Local dev environments
- Documentation
- Process automation

Works great if it's cohesive. Hodge-podge, not so much.

Are we designing accordingly?