Ala Shiban

Co-Founder @GetKlotho - Crafting the Adaptive Architectures paradigm. ex-Microsoft, ex-Riot Games

[Ala Shiban @AlaShiban]

@LawrenceDCodes So much of what is said is also purely wrong.

[Ala Shiban @AlaShiban]

@editingemily 'Your plane is now 99% landed , 5 minutes left, please wait'

[Ala Shiban @AlaShiban]

We need to make VPS's on AWS cool. Lightsail was the right idea, but it's so disconnected from the rest of the AWS ecosystem that it feels wrong to use it

[Ala Shiban @AlaShiban]

Noodle disagreeing on the role Kubernetes plays in the infra space, he wouldn't let it go

[Ala Shiban @AlaShiban]

amethyst is the app I can't run my Mac without anymore. Auto-tiling is so convenient. I just wish there were easier ways to describe layouts and how windows fit in

[Ala Shiban @AlaShiban]

@seldo The call is on GenAI, not AI. AI/ML have been super useful for tens of years, since OCRing mail addresses. LLMs/GenAI in their current form do not justify their cost - I think they're spot on, and the data is behind them. The web wasn't a narrow piece of tech, stretch to compare.

[Ala Shiban @AlaShiban]

If there's one thing I've learned from that relationship, it's that those two ends are hard to bridge 😅
3/3

[Ala Shiban @AlaShiban]

I had a partner with perfect memory recall (inherited genetic 'bug'). I don't have a strong memory. I found that my interests naturally gravitated towards abstraction and generalization, whereas hers gravitated words specialization and building more of the 'branching tree'.
1/3

[Ala Shiban @AlaShiban]

To me the truth was in the reduction into rule-sets, the patterns -- the way to discern how the underlying systems works. For her it was unnecaccery, it was all if-then statistics, she could see it all in her head.
2/3

[Ala Shiban @AlaShiban]

I remember how critical it was for me to learn to finish things. Starting was easy, finishing is hard, especially with ADHDish-ness.

[Ala Shiban @AlaShiban]

prediction: a big AI breakthrough will come from the interplay and integration of formal verification/logic provers tools like Lean 4.
LLMs will 'conceptualize' and formalize, provers will deduce and infer, then LLMs will read the results and update it's understanding of what's true.

[Ala Shiban @AlaShiban]

Here's how Klotho2.x will work

#InfrastructureAsCode #IaC #DevOps #CloudComputing #Terraform #AWS #Azure #GoogleCloud #Kubernetes #CloudNative #OpenTofu #CloudFormation #Crossplane #CICD

[Ala Shiban @AlaShiban]

@ylecun @ylecun What role do you see for Math provers and logic solvers in the context of a more capable AI system? Analogy wise feels like they play more of a left-brain computation that we're missing?

[Ala Shiban @AlaShiban]

@buckymoore Congrats @buckymoore !

[Ala Shiban @AlaShiban]

infra and cloud approaches 2024 edition #InfrastructureAsCode #IaC #DevOps #CloudComputing #Terraform #AWS #Azure #GoogleCloud #Kubernetes #CloudNative #OpenTofu #CloudFormation #Crossplane #CICD

[Ala Shiban @AlaShiban]

infra and cloud approaches 2024 edition. @davidfowl Aspire would be under AppArch right? #InfrastructureAsCode #IaC #DevOps #CloudComputing #Terraform #AWS #Azure #GoogleCloud #Kubernetes #CloudNative #OpenTofu #CloudFormation #Crossplane #CICD

[Ala Shiban @AlaShiban]

@davidfowl the difference is iteration I think. I think with code-gen you can't reasonably take in the code you genned, edit it, and reasonably allow the original system to continue iterating on the same code, whereas with LLMs you can. (to the degree that they do what they do)

[Ala Shiban @AlaShiban]

TLDR of Infra & Cloud Tooling Approaches in 2024:
1. Terraform*
2. Terraform as programming language code
3. Terraform apply in a loop running as a Kubernetes operator
4. High level application architecture compiled down to Terraform
5. Combining application architecture and application code into 1
6. Auto-generated terraform based on application code

blog coming soon

* Terraform the idea, not the actual implementation
#InfrastructureAsCode #DevOps #CloudComputing #terraform #Kubernetes

[Ala Shiban @AlaShiban]

friend asked me about pitch decks recently so I re-shared this side project I built with 15k searchable decks: https://t.co/by6UFjxpAr

[Ala Shiban @AlaShiban]

We need one of the always-on container services on @awscloud that spins up/down in seconds and doesn't have a cluster fee. 23 minutes to get one up and running is way too long