Jan 1, 2025

How we built The SaaS product with microVm

Initial Struggle after Graduation

After finishing my internship and completing college around June 2024, I entered the “apply → wait → rejection” loop. I was aiming for DevOps roles, but every job description demanded deep specialization, certifications, and industry-level experience. With no real industry exposure, almost no one shortlisted me.

But I kept studying — AWS, Proxmox, LXC containers, and Linux internals. At the same time, I was procrastinating applying for jobs because somewhere inside, I felt I wasn’t “ready enough.”

Before I realized it, six months had passed. It was January 2025. A new year… and I was still jobless.

Frustration peaked.

That’s when I told myself:

“Forget it. I’ll build my own mini-AWS. A full project. Something I can actually show recruiters. And if no company hires me… I’ll make this real myself.”

Redefining the Dream

I started redesigning my old project idea and simultaneously went deeper into the real core tools every DevOps engineer needs — Linux, Terraform, Ansible, AWS, networking, everything.

I first thought of using Proxmox as the hypervisor. But I kept researching to see what the industry standard really looked like.

That’s when I discovered OpenStack and OpenShift.

And my reaction was literally:

“Damn… this is exactly what I need.”

So I spent the next month learning the fundamentals of both. My plan was ambitious:

  • Install OpenStack on my local PC.
  • Use its APIs for VM lifecycle, networking, EIPs, security groups, etc.
  • Create a VM inside OpenStack.
  • Install OpenShift on that VM for Kubernetes workloads.
  • Upgrade my Node.js API so it could talk to both OpenStack and OpenShift.

It was tough. Huge systems. Massive documentation. Errors everywhere — API failures, CloudInit issues, broken networking, you name it.

But after a lot of trial, error, and frustration… I managed to get a manual setup working.

Then I expanded the idea:

“Let’s integrate Kubernetes + Ansible as well, so in the future we can build a smooth CI/CD pipeline.”

And all of this brought me to May 2025.

Reality Check

While working on this massive architecture, I was also applying for jobs. But the same question kept hitting me:

“How will I even host this? A decent server costs at least ₹10 lakh. I’m building everything on my PC. I don’t even have proper internet to host real clients.”

No job.
No money.
Huge project.
No clear path forward.

So I paused the idea.

For the next two months, I focused only on the fundamentals — AWS, Terraform, Ansible, Linux, Git… and kept applying for jobs.

Still nothing.

By July 2024, the panic started again.

The Turning Point — Firecracker

Around this time, I cleared the Solutions Architect Associate exam. This gave me a big confidence boost.

I still wanted to build a mini-AWS. This dream had been with me for almost 2–3 years.

Then I had a crazy thought:

“What if I host it on AWS itself? But AWS is so expensive… I need something cheap.”

While researching AWS Lambda, I had a realization:

AWS obviously doesn’t run Lambda functions inside Docker containers directly on bare servers — security would be a nightmare.

So how are they running and destroying environments so fast?

That’s when I learned about Firecracker — Amazon’s open-source microVM technology built specifically for Lambda.

Mind blown.
I loved the concept instantly.

Firecracker creates super lightweight VMs with their own stripped-down OS + kernel. Inside those VMs, AWS runs user workloads securely in isolation.

Then I discovered that many companies were using Firecracker + containerd to give users isolated Docker environments safely — because each Docker runs inside its own microVM.

Exaple platforms : Weave Ignite, Fly.io, Coder, etc.

And then my idea clicked:

“Where everyone else gives containers inside microVMs… I will give microVMs themselves to users.”

They are cheap, extremely lightweight, secure, fast to create/destroy — perfect for a cloud product.

Then I found Flintlock, a tool for managing Firecracker microVMs in a cloud-native way.

And everything finally made sense.

Bringing It All Together

So I built the architecture:

  • Flintlock for provisioning microVMs
  • containerd + Firecracker as the VM runtime
  • A custom API FastAPI to manage everything
  • Frontend connected to our microVM orchestration API
  • Terraform for provisioning EC2 hosts where Flintlock runs
  • Ansible for updates, patches, and configuration
  • K3s integrated for lightweight container orchestration
  • Everything deployed on AWS, clean, automated, and scalable

And that system became our SaaS product — the mini cloud platform I had always wanted to build.

From being jobless and frustrated…
to building something I once thought only big companies like AWS could build…

This is how our product was born. 🔥