My Evolving Workflow as a Software Developer

2026-04-19

I have been in the software development space for roughly 15 years and counting: from my first "Hello world!" print statement in Visual Basic as a student, to setting up automated systems and complex tasks programmatically now. As a former avid blogger and flash fiction writer back then, I would have hardly believed a word you said if you told me I would be crafting more logically coherent English prompts than typing C# or Python syntax. Yet this is a looming reality.

At the Anderson School of Management in ABQ, I made it a point to understand the fundamentals of model training and optimization in my Data Mining class, as well as what it took to create simple Python pipelines to tokenize texts for my Text Mining and Sentiment Analysis class. This was prior to the pandemic years. My priority was to learn enough to get a job that paid the bills and kept me mentally engaged and excited. I leaned towards programming for implementation and not necessarily the concepts, totally unaware that several labs across the globe were cooking what would change the programming playing field—RAG, LLMs, and Generative "Artificial Intelligence."

I have always looked to blogs, code repositories, Stack Overflow, and documentation sites to bootstrap my knowledge on how to implement custom features, connect systems, and cross-check syntax and best practices. Then came the programming tools that promised to reduce my research overhead when implementing a new task or greenfield project. Initially flawed in many ways, and progressively improving in complexity and accuracy at an overwhelming rate, AI tools and their developers have disrupted our workflows and daily experiences without our consent, and it is not stopping anytime soon.

As with natural selection, you adapt or stay stubborn and arrive at the place where you are still trying to improve car speed while the rest of the world is optimizing hyperspeed trains and teleportation. I felt a lot of guilt initially when I started testing out the capabilities of some of the most popular AI coding tools. I felt that my code was my work and that if I relied on the assistance of an AI agent in coding, I was cheating and doing something illegal.

As I mature both in my career and person, I quickly realized that I was harboring such a useless emotion. My code does not matter! My gradual mental shift from a product or service engineer to an engineer of experiences and value opened up the synapses that accelerated my developer experience and efficiency.

I learn way more now and forget even more with just the sheer amount of information, options, and decisions that I have to deliberate on to perform meaningful programming tasks with AI. I have become more of a manager of software development than a software developer. Over the past two years, I noticed personal growth as a strategist who prioritizes higher-level architectural decisions and analyzes trade-offs. And I understand that succeeding and finding fulfillment in software engineering will eventually transcend the dopamine hits from solving an impossible algorithm by hand, solving some painful bug or refactoring an entire codebase.

I'm not entirely sure what the future will look like with the current trend, but I will still choose to focus on delivering human value beyond expectations. To do this, AI and code are but tools.