A recent blog post that resonated with me: The Year Coding Sped Up 100x:
The point is not that software engineering is dead.
Quite the opposite: the engineering in software engineering is thriving. Modeling, formal validation, higher-level languages—these were at the fringes of "traditional" software engineering for a long time because most of the brain power was spent writing code.
As a junior who started coding in 2023, I can deeply relate to that statement. This is the story of how AI tools transformed me from a struggling developer into a software engineer.
The Beginning: Learning the Hard Way (2022-2023)
When I started "learning code" at school in 2022, it was truly learning the hard way. Nothing comparable to formal textbooks—I had to grind through exercises, maintain a personal dictionary of vocabulary I struggled to remember, and piece together understanding bit by bit. At that time, GPT-3 was just starting out, mainly a playground for people asking it to tell jokes.
For a year, I took on the developer role in school projects because I wanted to learn more. I spent hours checking other people's code to copy-paste parts of it, reverse-engineering solutions without fully understanding them.
The Chile Semester: Coding in Three Languages
Then I spent the first semester of 2023 in Chile, taking classes like "web development fundamentals," "databases," and "algorithms." The real names were "Inicio Al Desarrollo Web," "Base de datos," and "Algoritmos"—all classes taught in Spanish, code in English, and my brain thinking in French. I couldn't describe this mess better than: "In my code, some comment lines were written in 3 languages."
During these classes, we worked with React and Koa, SQL, and C++, respectively. They were all supposed to be entry-level, but the students around me had already completed 2-3 semesters of similar courses and were far more advanced than me. I kept asking to see others' code, spent countless hours reading the easier classes' slides first to make my way through understanding the actual course material I was supposed to be following.
First Internship: Contradictions and Crisis (October 2023)
In October 2023, I started my first internship. GPT-3 was being fine-tuned to become what we now know as modern AI. At that time, I experimented a bit out of curiosity—asking it to explain pieces of code, which I would confirm with the classic StackOverflow questions and other sources. I also used it for repetitive tasks, like extracting just the keys from JSON into text format. Answers started to make sense. Others were using GitHub Copilot, which I didn't like because it was writing the code for me without giving me time to think about what I was doing.
The Identity Crisis
When I started my internship, I was full of contradictions: both happy to learn the new languages I was coding with (which I love to learn) and disappointed seeing people with such advanced skills that would take me years, maybe decades, to acquire. There I started questioning my career as a developer (vocabulary matters here).
In the meantime, the use of AI was rising and we kept hearing people say we would lose our jobs. What was I doing in a world where I was just a beginner, facing senior people with much more knowledge than me, and facing new AI tools that were becoming as efficient as an intern?
Back to Fundamentals (Fall 2024)
A year later, I returned to school to study "Fundamentals of Computer Science." I had briefly touched some C++ code before, but I was really lacking foundational knowledge. We learned Rust, which, coming from Python, really kicked my ass. We did Assembly and Haskell, and kept listening to our teachers who were so enthusiastic about low-level languages and optimizations.
I can painfully tell you that GPT had not been trained on Assembly books yet...
This was a great time—stepping back from speed-coding and "AI taking over the world," and trying to redefine what "software engineer" meant for us. It grounded me in the fundamentals and gave me perspective on what engineering actually entails.
The Turning Point: Datadog (Spring 2025)
After that semester, I started my internship at Datadog in New York. I was in a great team with excellent colleagues who taught me a lot. I started learning Go, using AI the best I could to accelerate my learning—helping me ship code while also teaching me core principles of the language. I had an excellent mentor who did really strict PR reviews that helped tremendously with my learning.
Innovation Week: AI Everywhere
One month in, we held an innovation week, trying to put AI everywhere in our product. At that time, all companies were releasing AI assistants—we were building Bits AI, others had their own solutions. The race had started.
In the following month, I started working on an NLQ (Natural Language Query) feature for which I had to read a lot of blog posts, study how others were approaching the problem, and experiment extensively. I learned a lot about how to work with AI and get these capricious little minds to respect what I was asking them to do.
The Hackathon Week: Meeting Cursor
After this period came a company-wide AI hackathon week, during which we attended numerous conferences and gathered for vibe-coding sessions where we built apps in an hour. At that time, I was pretty skeptical about AI, given my recent battles with it. I finally accepted to install Cursor (I didn't even have Copilot in VSCode as I didn't like the noise it brought).
And Cursor suddenly became my best colleague.
I was suddenly able to debug my code 10x faster, delegate boring tasks to AI, and read documentation or answer messages in the meantime. This was the moment everything changed.
My Current Setup
Let's jump to what my setup and workflow are now: I run multiple Claude Code instances simultaneously to handle most of my AI tasks, combined with Cursor, which helps me read my code as an enhanced IDE. I use it to ask for clarifications on specific bits of code, no longer than a few lines.
Redefining the Role: Developer vs. Software Engineer
In a world where AI is, for some, a confidant to which they tell everything and ask to make important decisions in their lives, or use to justify bar discussions with a quick prompt, AI can also genuinely enhance some of our abilities.
I've had many interesting exchanges with quite a few engineers in the company now, which helped me gain perspective (I also had a month off between my internship and my full-time job—take holidays, devs!).
What "Software Engineer" Actually Means
And now I understand more clearly what my job is: I am a Software Engineer. My job requires developer skills—I need to be able to read code, understand how it works, know best practices in several different languages, and know how to deploy a service.
But as seniors were before me, I am more of an orchestrator than a code writer. My job is not to write lines of code. I can now respond to support messages, write fixes, deploy them, make sure they run in production, and manage the entire lifecycle of a feature.
The Transformation
As a junior, every part was scary, and the "write the fix" part was perhaps the worst: I could see myself fighting with language syntax for countless hours, reading again and again my personal notes on the language, realizing which patterns didn't actually apply in every situation, going back and forth with StackOverflow...
Now, every time I have to write code, I can stop worrying about that part. I focus more on:
- Defining my needs clearly
- Architecting solutions that AI will implement for me
- Writing tests that aren't influenced by the code I wrote
- Collaborating with other engineers and teams
- Organizing projects end-to-end
- Ensuring quality and maintainability
Altogether, this makes me more and more an Engineer, not just someone who writes code.
Conclusion
The transformation from developer to software engineer isn't about abandoning coding—it's about elevating your role. AI tools like Cursor and Claude Code haven't replaced me; they've liberated me to focus on what truly matters: engineering decisions, system design, collaboration, and delivering value.
The fear I felt in 2023—that AI would make me obsolete before I even started—was misplaced. Instead, AI accelerated my journey from a struggling beginner to a being-confident engineer who can architect, ship, and maintain production systems.
The engineering in software engineering is indeed thriving. And I'm thrilled to be part of it.
Top comments (0)