So I Started a Cloud Company Out of Spite
How Light Cloud Was Born from 14 Hours of Kubernetes Debugging
Remember that 14-hour Kubernetes debugging session I mentioned? The one where I fixed a race condition that AWS's own documentation said "shouldn't happen"? Well, after implementing the fix that three different AWS architects couldn't figure out, I thought I'd won.
Then our monthly AWS bill arrived: $4,847.
Not because we did anything wrong. We followed every "best practice." We used their recommended architecture. Auto-scaling groups? Check. Multi-AZ deployment? Check. Read replicas? Check. We built it exactly how the AWS Solutions Architect would.
The app was handling 10k requests per day. The infra could handle 10 million. We were paying for a Ferrari to deliver pizza.
That's when I snapped. Not like a normal person who might take up yoga or start a garden. No, I did what any sleep-deprived, caffeine-addicted developer would do: I decided to start a cloud infrastructure company.
My therapist says it was a "trauma response." I prefer to call it "entrepreneurship."
The Exact Moment Light Cloud Was Born
Picture this: It's 4 AM. I'm sitting in my underwear, surrounded by empty energy drink cans, staring at a CloudFormation template that's 1,847 lines long.
I turned to my rubber duck (yes, I debug with a rubber duck, don't judge) and said: "What if infrastructure just... worked?"
The duck said nothing, which I took as agreement.
That night, I wrote a manifesto on a napkin. Well, actually it was on my iPad, but "napkin" sounds more startup-y. It said:
INFRASTRUCTURE SHOULD BE:
- Like LEGO blocks..
- Not require a PhD in Computer Science
- Just F***ing Work
- Cost less than a car payment
- Deploy in under 2 minutes
Two weeks later, Light Cloud, Inc. was... well, not born exactly. More like conceived. In a very preliminary, "this might be a terrible idea but let's try it anyway" kind of way.
Finding My Co-Founder (Or: How Google Meet Became Our Office)
You can't start a company alone. Well, you can, but then who do you blame when things go wrong?
Alexis and I had been friends for years. She's a non-technical entrepreneur in California who'd watched me descend into infrastructure madness from 9 time zones away. Our friendship survived on WhatsApp rants and video calls where I'd explain why I was still awake at 3 AM fighting with AWS.
Then one day, she called me. "Julia, I just got a $18,000 quote from a software agency to deploy my platform. They're saying it needs something called Load balancer, Kubernetes and.. I forgot. Isn't that the thing you're cursing at?"
She went through three development agencies already. Each one spoke a different language... microservices, serverless, container orchestration. All she wanted was for her clients to log in and use her platform.
Over a very pixelated Google Meet call (because of course my internet chose that moment to act up), I showed her my napkin manifesto about infrastructure blocks. She didn't understand the technical details-and that was exactly the point.
"Wait," she said, her video freezing mid-gesture. "You're telling me this could make it so normal people like me don't have to learn what a load balancer is?"
"That's the idea."
"Julia! We need to build this. You know the technical nightmare, I know how to be a human. (she said while laughing) We're starting a company."
"We? You're in California. I'm in Poland. How is that going to work?"
"It's 2025. Everything is remote. And let's be honest-I'll be the one talking to humans because you explain everything in binary."
She wasn't wrong. My idea of small talk is discussing database indexing strategies. Alexis can actually have a conversation without mentioning Git commits. It's a superpower I don't possess.
And that's how my best friend became my co-founder. Our "headquarters" is a Confluence workspace. Our meeting schedule is a constant negotiation between PST and CET. I code while she sleeps. She takes calls while I sleep. It's like a relay race.
The MVP That Actually Works
Here's what nobody tells you about building a cloud platform: you need... a cloud platform to build your cloud platform. It's like needing experience to get a job that gives you experience. It's turtles all the way down.
I spent two weeks building our first proof of concept. Then another month turning it into an actual demo that doesn't crash when you look at it wrong. We can now deploy a real app with a database in few minutes. It works with Node.js and even PHP (because someone has to love PHP).
Is it production-ready? God no. But it works well enough that we can show it to people without praying to the demo gods.
The breakthrough came when we showed it to a CTO who's also a cloud architect-the kind of person who's forgotten more about AWS than I'll ever know. We were expecting her to tear it apart.
Instead, she sent us a message: "I got you ;) finally a way to put the personal brand at work for a cool project! I absolutely love the name of the company and the concept you are trying to build. I really believe it has potential!!!"
When someone who's actually built infrastructure at scale wants to be your advisor just because she believes in the project? That's when we knew we might not be completely delusional.
Having her in our corner-someone who gets both the technical complexity and sees why our simple approach could work? That changed everything.
Well, mostly not delusional.
Why "Light Cloud"? (The Naming Marathon)
Naming a company over video calls with a 9-hour time difference is special kind of hell.
We went through dozens of names:
- CloudSnap (taken)
- EasyDeploy (sounds like a disease)
- ServerLess (literally the opposite of what we do)
- InfraBlocks (meh)
- YAMLBegone (Alexis couldn't stop laughing)
- DeployBoy (just... no)
After weeks of this torture, at 2 AM my time, 5 PM her time, both of us exhausted for different reasons, I had a moment of clarity: "We just want to make the cloud lighter. Like, less heavy. Light... Cloud?"
Alexis went quiet for a second, then started laughing. "That's either brilliant or we're both too tired to think straight. Wait—it works on multiple levels. Light as in not heavy, light as in illuminating the black box of infrastructure, light as in fast..."
We sat in video call silence for 30 seconds (which feels like 3 hours on a call). Then I checked the domain. Available.
"That's a sign," I said.
"Or we're both too tired to think straight," she replied.
We bought it anyway. The domain alone cost us $1,200. That's when we learned our first lesson about starting a company: everything costs more than you think.
What's Next?
Look, I know what you're thinking. Two people, no funding, competing with tech giants-what could possibly go wrong?
Everything. Everything could go wrong.
But here's the thing: every big company started with someone stupid enough to think they could fix a problem everyone else accepted as "just how things are." AWS started because Amazon needed better infrastructure. Stripe started because two brothers were tired of payment integration hell.
We're starting Light Cloud because we're tired of infrastructure complexity hell.
In six months, we might be another failed startup statistic. Or we might be the reason you deployed your last app without writing a single line of YAML. We might be the reason a non-technical founder finally launched their idea without spending their kid's college fund on AWS consultants.
The prototype is rough. The business plan is basically non-existent. We're running on caffeine, spite, and the stubborn belief that infrastructure doesn't have to be this hard.
But we have something AWS doesn't: we remember what it's like to be a developer who just wants to ship code. We remember the frustration, the 3 AM debugging sessions, the moment you realize your AWS bill is higher than your salary.
We're building Light Cloud for every developer who's ever wanted to throw their laptop out the window after reading Kubernetes documentation. For every founder who's been quoted the GDP of a small country just to deploy a simple app. For everyone who thinks "there has to be a better way."
There is. We're building it. It's held together with shell scripts and prayers right now, but we're building it.
Join us on this ridiculous journey. Sign up for our waitlist at light-cloud.com. Follow our progress (or spectacular failure) as we try to make infrastructure boring.
Because if two sleep-deprived people working across 9 time zones can make deployment simple, imagine what we could do with actual resources. Imagine a world where "deploy to production" doesn't trigger PTSD. Imagine never having to explain to a client why hosting their simple website costs more than their car payment.
That's the world we're building. One infrastructure block at a time.
Will we pull it off? Come back in six months and find out.
But I'll tell you this: we're going to give it everything we've got. Because the alternative-accepting that this is just how infrastructure works, is unacceptable.
The cloud should be light. Simple. Accessible.
And maybe, just maybe, what this industry needs is exactly what we're building-a fresh perspective from people who understand both the technical complexity and the business reality of modern software deployment.
The learning curve has been vertical. Incorporating a company across two countries? Stock options when you have no idea what your company is worth? Every day brings a new thing we didn't know we didn't know.
But that's what makes this exciting. We're not just building a product-we're learning how to build a company. And if we can figure out Delaware C-Corps and vesting schedules, we can definitely figure out how to make infrastructure simple.
P.S. - Seriously though, if you know anyone who's as frustrated with cloud complexity as we are, send them our way. We're building a community of people who believe infrastructure should just work. Even if we fail, at least we'll fail together.