º£½ÇÂÒÂ×ÉçÇø

The widespread adoption of Large Language Models (LLMs) has transformed the software engineering landscape in ways that require reevaluating workforce education needs. Educators are increasingly curious about the relevance of teaching students how to code in this new era. However, we lack evidence of how software engineers themselves perceive this shift. We report results from a study exploring what engineers think about learning to code today and how students should use LLMs to learn to code. We highlight the 'fundamental skills' software engineers believe students need to master, even though LLMs might be good at doing them. We conducted in-depth interviews with software engineers from the United States and Nigeria between August 2024 and January 2025 to create a current and inclusive report across diverse socioeconomic contexts. We used LLMs to support our thematic analysis. We used Latent Dirichlet Allocation(LDA) and BERTopic models to capture additional insights. Our hypothesis was that 2 years post-LLMs, there would be a significant change in the skill requirements for software engineers, which in turn will influence what students need to learn. We found that most of the core skills for software engineering are still relevant. However, they are now augmented by AI for a faster and more seamless software development process.