Got laid off last week with the rest of the dev team, except one full stack Laravel dev. Investors money drying up, and the clowns can't figure out how to sell what we have.

I was all of devops and cloud infra. Had a nice k8s cluster, all terraform and gitops. The only dev left is being asked to migrate all of it to Laravel forge. 7 ML microservices, monolith web app, hashicorp vault, perfect, mlflow, kubecost, rancher, some other random services.

The genius asked the dev to move everything to a single aws account and deploy publicly with Laravel forge... While adding more features. The VP of engineering just finished his 3 year plan for the 5 months of runway they have left.

I already have another job offer for 50k more a year. I'm out of here!

I am glad that I have no idea what most of those terms are.

@aviophile to each their own. I love this stuff!

@lungdart i am more of a low level guy. Much more stable “stack”, no framework, just proper languages.

@aviophile it can be fun doing everything from scratch. Your velocity will suffer greatly, and you'll be maintaining everything yourself. Not a fun time.

Good for keeping skills sharp though

@lungdart "...your velocity will suffer greatly..."

Nah. There's nothing slower than a dev (or, more often and of necessity, a dev *team*) using all the frameworks and tools available.

@Noook a simple crud app of course!

A CTO came in and was worried about scale (even though we only had one customer and maybe dozens of transactions a month). He wanted rancher because he heard it was a good scaling tool but had no idea what it was (kubernetes cluster orchestration).

At the same time, potential new partners were looking for us to be SOC2 compliant and have SLAs of 3 9s.

So I built a highly available, distributed, SOC2 compliant infrastructure on EKS (AWS kubernetes service). By the time it was done, the potential partnership fell through but we already had the infrastructure... It cost about $3000/mth, but we had credits up the wazoo anyway...

Anyway they don't know how to work with it now. I was surprised they laid me off effective immediately instead of offboarding and knowledge transfer.

@spongegeoff there's a big difference between using a stack to reduce cognitive load and time to release, and using every new and shiny thing you see.

We went with a k8s stack with self hosted services for security compliance and SLA requirements.

@spongegeoff our team were releasing to prod several times a day with this stack. Some user stories took an hour from in progress to being in front of customers.

@lungdart Glad you're off to a better place. Sounds as if your previous management really struggle with their decision making.

@lungdart who talked about doing from scratch? Don’t tell me your only idea about embedded is reimplementing same protocols again and again.

If you try to throw a shade, do it in an informed way ;)

@aviophile #import <1wire.h>

I used to have a ton of homebrew embedded libraries back when I did embedded. These days what I can find on GitHub are faster, easier, and most importantly, someone else keeps it up to date with bug fixes, security patches, and language versions.

@lungdart damn that's not a bad stack! What a waste of it though...

Like building an entire giga factory for just one pie.

You're goals man. I'm doing MLOps here and Im looking up to you. Using Docker Swarm though and a lot of fancy stuff that exists on Kubernetes simply isn't available for the Docker Swarm ecosystem like ArgoCD

Started with a two node on prem infrastructure as that is all the client would be willing to invest.

Moving on to a bigger company where Kubernetes and GitOps are the norm though