When I first decided to learn Python, I quickly ran into a wall that a lot of aspiring developers face: time and money.
Being based in the UK with strict time restrictions, I couldn't afford a top-tier bootcamp or college course. On top of that, my local colleges simply weren't offering anything focused on Python development. So, I took the only route available to me: I became a self-directed learner.
Things have been going really well, but my learning path has taken some unexpected turns. Here is what I’ve learned about navigating the modern junior developer landscape.
Building the Foundation (The Hard Way)
In the beginning, it was just me, a code editor, and the internet. I lived in books, practically memorised YouTube tutorials, and scoured GitHub repositories to see how real applications were structured.
I’m still fully committed to progressing in manual coding. Knowing how to write for loops, build functions, and structure classes is the bedrock of everything we do. But as I started looking around at this ever-evolving industry, I had a massive realisation: just understanding and being able to write manual code is no longer enough.
My New Pair Programmer: AI
Because I didn't have a senior developer sitting next to me, I started leveraging AI. But I didn't just use it to generate code snippets. I started treating AI as my:
Teacher & Mentor: Breaking down complex computer science concepts.
Debugger: Helping me spot the missing colon or logical error that had me stuck for hours.
Friend / Pair Developer: Bouncing architectural ideas back and forth before writing a single line of code.
Using AI this way opened my eyes to where the industry is heading.
The Epiphany: Welcome to the "Vibe Coding" Era
If you’ve been hanging around dev communities lately, you’ve probably heard the term "vibe coding." It’s the idea that developers are increasingly prompting, testing, and iterating alongside AI to build software, rather than typing out every single line by hand.
I feel like this has almost become a mandatory part of the job. Companies hiring junior developers today want more than just someone who knows syntax. They want someone who can orchestrate AI. The career path for a junior coder has fundamentally shifted.
Levelling Up: Agents, Skills, and MCP Servers
Because of this shift, I decided to dive headfirst into AI engineering concepts to complement my Python skills. Right now, I'm doing extra study on:
AI Agents & Skills: How to build agents, give them specific tools (skills), and get them to work autonomously on complex tasks.
MCP (Model Context Protocol) Servers: Learning how MCPs securely connect AI models to external tools, file systems, and databases to give them real-world context.
Local vs. Cloud Models: I’m actively testing workflows on cloud-based LLMs as well as local models to understand the tradeoffs in privacy, speed, and capability.
OpenCode Environment: I'm using OpenCode (both the Desktop app and the CLI) to experiment with these agentic workflows in real-time.
My Thoughts
The barrier to entry is changing. The floor isn't just "can you write a function?" anymore; it's "can you read, review, and direct an AI to build a scalable system?"
I'm incredibly grateful for the manual foundation I’ve built, because without it, vibe coding just turns into blind copy-pasting. But by combining my self-taught Python fundamentals with learning a deep understanding of Agents and MCP servers, I finally feel like I'm not just learning to code; I'm learning how to build software in 2026.
Are there any other self-taught devs out there balancing manual coding with AI agents? I’d love to hear what tools you are testing in the comments!
Top comments (0)