Many of us are learning new tech are doing it for one reason—to get a new job. That’s valid, but it completely changes how you study, often leading straight into tutorial hell and forgotten fundamentals. When I set out to map 90DaysOfDevOps, and now while tackling data engineering, my baseline was different. I wanted to build, break, and understand the mechanics simply because the space interested me. Whether you’re trying to pivot careers or just add a new superpower to your home lab, here is the blueprint I follow to actually make new technologies stick.
Today, this challenge is amplified even more by AI. We live in an era where an LLM can spin up a Docker compose file, write a complex SQL query, or draft an entire CI/CD pipeline in seconds. It’s tempting to use AI to bypass the friction entirely. But if the AI does all the thinking, you get the illusion of progress without the muscle memory. The trick isn’t using AI to avoid learning—it’s using AI as an interactive co-pilot to explain error logs, challenge your assumptions, and accelerate your hands-on loops. Be smart with AI not addicted! Use it as a tool.
Start with Clear Goals (Treat Tech as a Tool, Not a Collection)
When you first look at a massive landscape like DevOps, Data Engineering, or Cloud-Native architecture, the sheer volume of tools can cause instant paralysis.
- Define the Scope: Break the massive umbrella term down into discrete domains (e.g., version control, CI/CD, infrastructure as code, monitoring).
- Audit Your Baseline: Honestly map out what you already know. Don’t start from absolute zero if you already understand networking or basic scripting.
- Anchor Your “Why”: Are you trying to pivot careers, solve a specific bottleneck in your current role, or just scratch a technical itch? Your motivation dictates the depth required.
- Write the Roadmap: Plan your milestones before you touch a keyboard so you have a finish line in sight.
Theory vs. Application (The 80/20 Hands-On Rule)
We have all accumulated digital graveyards of half-finished Udemy courses and O’Reilly books. Passive reading feels like progress, but it’s an illusion.
- Resist Tutorial Hell: Limit the time spent watching videos or reading docs to the absolute minimum needed to understand why something works.
- Embrace the Friction: No cheat codes or blindly copy-pasting GitHub snippets. Type out the code, break the config files, and stare at the error logs—that is where the real learning happens.
- Find Your Medium: Know how you best absorb foundational concepts (whether it’s technical documentation, video walk-throughs, or audiobooks on a commute), but always pivot quickly back to the terminal.
Solve Real, Relatable Problems
Abstract tutorials are boring, and they don’t stick. The best way to learn a technology is to force it to solve a problem you actually care about.
- Build a Home Lab: Spin up a mini rack, deploy a Kubernetes cluster, or automate a painful manual chore around your house or workflow. I have written a lot about homelabs here, they don’t have to be expensive, especially today where it seems components are around 10x more than they were a year ago! A laptop or even a VM on your machine is something to play with.
- Scratch Your Own Itch: Build something useful—even if a commercial tool already exists. The act of engineering the solution teaches you the underlying mechanics. I have tons of ideas that pop into my head, they are not always niche ideas but they are great to just build a prototype, understand what is happening and adapt, improve or delete it and move on.
- Make It Interesting: If you are bored by the sample project (like yet another generic to-do app), change the domain to something you love, whether that’s tracking video game stats or analysing sports data. I am currently learning lots about data engineering and using my cricket team stats and data to help keep me interested.
Incremental Learning & Feedback Loops
Learning isn’t a linear climb; it’s a series of iterations. Once you get something working, the real question is: Could it be better?
- Iterative Refactoring: Get the MVP working first. Once it’s functional, go back and optimize it for security, efficiency, or scalability.
- Embrace the Error Log: Treat error messages not as failures, but as your primary feedback mechanism. Debugging teaches you more in ten minutes than a smooth tutorial does in ten hours. “Learn from your mistakes”, why did this fail?
- Build in Public: Share your code, document your journey, or write about what broke and how you fixed it. Explaining it to others forces absolute clarity. I appreciate not everyone is a blogger or youtuber but it’s a great way to build accountability, during 90daysofdevops I was tweeting updates every single day and I know people were getting annoyed “Just let us know if you complete it and will go back and read it then” but me putting 10/90 out there was my own accountability to finish that long list and learn.
Consistent and Correct Action
Occasional weekend binges won’t build durable neural pathways. Technology moves too fast for sporadic cramming. I can sit on YouTube all day watching videos about relatable technical videos that are associated with the plan, but sometimes that is just background noise and its not solid learning time. You have to have that right mindset to be in learning mode.
- Compound Your Efforts: Thirty minutes of focused, hands-on practice every single day beats a five-hour marathon session once a week.
- Practice the Right Habits: “Correct” action means not taking shortcuts early on. Learn the proper syntax, understand security best practices, and use version control from day one so you don’t have to unlearn bad habits later.
I am no learning oracle by any stretch, I was terrible in school. I have just been able to pick up topics I am interested in, I am equally in a privileged world where I have the time to spend time and learn new areas without the forcing function of finding a job or being desperate for that move into this new area.
As with all the posts, sometimes I write stuff down for me to go back to and remember the why. But as always if this helps one person get a plan together and execute on that then it is worth posting on this little corner of the internet for all the AI monsters to find.