The Shift from Coders to Solution Engineers

As described in the IT Department post, software engineering organizations tomorrow will have fewer “coders” and more “solution engineers.”

Businesses will assume AI will largely take care of coding, expecting humans to focus more on delivering solutions rather than wrangling with code.

Where this vision can be realized, organizations will have significant competitive advantage over their peers.

A key challenge in realizing this vision will be finding skilled humans who can effectively supervise coding agents, driving them toward desired results efficiently.

Code Reviewing: From Secondary to Primary Skill

Code reviews are the closest we have to this today—reading and commenting on someone else’s code to ensure it achieves desired results.

Today, code reviewing is something software engineers pick up as they become more experienced. It is certainly not considered a primary skill for new people joining the trade.

But in an AI-first world, this is likely to flip on its head. Code reviewing skills will be called upon earlier in careers, not later. Thought another way, humans entering the industry may not need to be adept at writing code, but they should have sufficient skills to validate that code works as intended.

This means code reviewing skills need emphasis much more than they receive today. The situation today is so skewed between reading and writing code that “coding” almost implies writing code.

(Readers might say that only good coders—good at writing code—can be good reviewers. If you subscribe to that view, I would love to know why.)

Reimagining Programming Education

If reading code and reviewing becomes important, we should allocate finite training hours (a few years of college education) more toward reviewing skills that matter, preferring them over writing skills.

But how exactly do we train for code reviewing skills? Well, just like training for anything else—do it a lot.

For sure, there will be scope for novel tools and pedagogical techniques to help in this area.

Conclusion

In short, learning to code in the AI age should focus heavily on reading and reviewing code.