I recently watched a video about the difference between senior and junior software engineers by Continuous Integration.
If I try to boil it down, the main difference lies in the ability to solve problems. Some of the skills needed for this can only be gained through experience, which is something I completely agree with.
This got me thinking about the early debates during the generative AI boom, where people insisted that AI could never replace human creativity. However, with my limited experience (I don’t think I will ever attain the legendary status of “Prompt Engineer” ¯\_(ツ)_/¯) using generative AI tools for coding, writing, and image generation, those tools seem “creative” enough to me.
A 2024 paper mentioned that:
“We theoretically proved that AI can be as creative as humans, provided it can effectively fit conditional data… The challenge lies in the availability of sufficient conditional data satisfying the desired level of creativity and the capacity of the AI model to fit those data.”
But knowing how these language models (LLMs) work—at least at a high level, it’s clear they don’t really “create” anything from scratch. Instead, they pull, randomize, and mix data from their “learning experience.” Even the paper emphasizes the importance of data in making AI appear more “creative.”
What does this have to do with being a senior engineer?
The more senior an engineer is, the more they are expected to solve bigger and more complex problems. Rather than chasing the glorified idea of “thinking out of the box” or magically conjuring creative solutions, I believe expanding your breadth and depth of knowledge and experience is far more attainable.
AI looks creative in problem-solving because it has access to vast amounts of data humans have already generated. You can, too. The more you learn and experience, the better you’ll be at solving problems.
While learning new things like a new language (or even juggling!) is beneficial, don’t forget to deepen your expertise in a few key areas that matter. Aim for T-shaped skills—a broad base of general knowledge with deep expertise in one area—or even π-shaped skills, as Forbes suggests.
Recently, my colleagues and I faced a performance bottleneck caused by a customer-written plugin. The challenge? We couldn’t modify the plugin code and had to treat it as a black box.
Because I knew there were only five entry points to the plugin, I suggested duplicating the plugin for each entry point and measuring performance for each from our side. Long story short, we pinpointed the bottleneck.
Looking back, I wouldn’t have come up with that suggestion if I didn’t already understand how the plugin entry points worked. Plus, my past experience profiling code for similar issues played a critical role.
So, I didn’t “think out of the box.” I generated the solution using a mix of past learnings. Maybe I’m an AI?
anggorodewanto.com 
Leave a comment