The Uptime Engineer
👋 Hi, I am Yoshik Karnawat
You'll see the exact projects I'd build if I started today. No certifications, no 90-day roadmaps, just the sequence that gets you from zero to hired. You'll learn which tools matter (and which ones waste your time), how to prove you can ship infrastructure (not just talk about it), and how to stand out in interviews by showing real deployments instead of memorized answers. By the end, you'll stop asking "what should I learn?" and start shipping systems.
You've seen the roadmaps.
Half say "get AWS certified first"
The other half scream "master Kubernetes immediately"
Both miss the point.
I started as a PhonePe intern 2.5 years ago, barely knew what a k8s cluster was.
From managing edge infrastructure processing 72 million transactions daily to switching over to a low-code data-engineering startup.
If I had to restart today with zero experience, I'd ignore 80% of what LinkedIn tells you.
Here's why: certifications teach you to pass exams. Projects teach you to ship infrastructure.
The difference gets you hired.
Month 1-2: Linux + One Cloud + Manual Deployments
Pick AWS or GCP. Not both.
Deploy something simple that accepts HTTP requests and returns JSON.
Do it manually. EC2 instance, security groups, SSH keys, the whole painful setup.
No Terraform yet. No automation. Just raw infrastructure.
Why this matters:
When you eventually write Infrastructure as Code, you'll know what's actually happening.
Most people skip this step and cargo-cult Terraform modules they don't understand.
What to build:
URL shortener (nginx + Python/Go backend + systemd service)
Monitoring script that tracks your laptop's CPU and sends Slack alerts when it spikes
Backup automation that syncs project files to S3 every night
These aren't impressive.
But they teach you what "a server" actually means, something people with three certifications still don't grasp.
Month 3-4: Docker + CI/CD + Zero-Downtime Deploys
Take that URL shortener. Containerize it.
Write a Dockerfile. Push it to a registry. Deploy it using GitHub Actions.
Skip the Docker Compose tutorials with 8 services.
You don't need Postgres, Redis, and RabbitMQ to learn containers.
You need one service that deploys automatically on every git push.
What to ship:
Dockerized app with automated deployments
Rolling deployment script using health checks (zero downtime)
Basic monitoring. Prometheus + Grafana answering "is my service alive?"
This is where you separate from tutorial followers.
In interviews, you can say: "I built a CI/CD pipeline that deploys to production on merge" and demo it live.
That's not a buzzword. That's proof you ship.
Pro tip: If you want to save the humongous AWS/GCP cloud costs for your projects, I’d suggest you checkout Hetzner. Not affiliated to them, I have been personally using their $5 server to run my side-projects.
Month 5: Infrastructure as Code (Make It Repeatable)
Take everything you built manually in Month 1-2.
Now define it in Terraform.
Every EC2 instance. Every security group. Every S3 bucket.
Start with one resource. Then two. Then a complete environment you can destroy and rebuild in 5 minutes.
Focus on three things:
State management (local first, then remote in S3)
Modules (when to abstract, when copy-paste is fine)
Secrets management (how to pass credentials without hardcoding)
Don't learn Terraform by reading docs.
Take something you already deployed manually and translate it to code.
The pain points you hit? That's where real learning happens.
Month 6: Kubernetes Fundamentals (Not CKA Prep)
Everyone says "skip Kubernetes until you have 2 years of experience."
They're half-right.
Don't try to become a Kubernetes expert.
Do learn why it exists and what problems it solves.
Run a local cluster using Minikube or Multipass.
Deploy your containerized app. Add a LoadBalancer. Kill a pod and watch it restart.
Understand three concepts:
ReplicaSets (how many copies of your app to run)
Deployments (how to update without downtime)
Services (how traffic reaches your containers)
That's it.
What to build:
Multi-replica deployment with health checks
ConfigMap for environment variables (no hardcoded secrets)
CPU-based autoscaling (just to see it work)
You don't need StatefulSets, Operators, or Helm charts yet.
You need to understand: "My app runs in containers. Kubernetes keeps them alive and routes traffic."
When you can explain that sentence with real examples, you're ahead of most junior candidates.
What I'd Skip Completely
AWS Certifications (Before Shipping Real Projects)
Certifications teach you to memorize service names.
Real projects teach you to debug production outages at 3 AM.
Get certified after you've shipped systems—when the knowledge sticks because you've felt the pain.
Learning 5 Programming Languages
Pick Python or Go. Ship with it for 6 months.
One real project beats ten "intro to programming" courses.
Waiting Until You're "Ready" to Apply
You'll never feel ready enough.
Companies hire people who ship things.
Not people who completed roadmaps.
Build 3-4 solid projects. Write about them. Apply aggressively.
Start This Weekend
Pick one cloud provider. Create an account.
Deploy something that responds to HTTP requests.
Break it. Fix it. Document what you learned.
Automate one painful manual task this week. A Bash script, a GitHub Action, doesn't matter.
Post about it publicly. LinkedIn, blog, X, just make it searchable.
You don't need 6 months to start. You need one weekend to ship something real.
Then repeat.
The students who land DevOps roles in 2026 won't have the most certifications.
They'll have a GitHub history of shipped infrastructure and a trail of posts explaining what broke and how they fixed it.
Start messy. Ship broken things. Fix them in public.
That's how you become dangerous.
Join 1,000+ engineers learning DevOps the hard way
Every week, I share:
How I'd approach problems differently (real projects, real mistakes)
Career moves that actually work (not LinkedIn motivational posts)
Technical deep-dives that change how you think about infrastructure
No fluff. No roadmaps. Just what works when you're building real systems.
Thank you for supporting this newsletter. Have an amazing year ahead.
Y’all are the best.