In the Age of AI, Adopt a "Mentorship Model" for Junior Engineer Development

Now that AI agents handle routine coding tasks, the software industry faces an urgent need to fundamentally rethink how it develops junior engineers. Scott Hanselman, known as a developer advocate at Microsoft, offered one potential solution in a podcast interview with Michael Stiefel.

Historically, junior engineers have tackled tasks such as bug fixes, routine code writing, and code comprehension—essentially "grunt work." By repeating these tasks, junior developers internalized design principles and cultivated a sense of code quality. However, with the rapid proliferation of AI tools, many of these entry-level tasks are being automated. This creates a structural problem: junior engineers are losing the very "hands-on learning opportunities" that once propelled their growth.

In his interview, Hanselman highlighted a model used in nursing called "preceptorship." This system pairs an experienced nurse with a novice nurse in a one-on-one relationship, directly transferring knowledge and clinical judgment through real-world situations. Critically, it is not merely "observing from the sidelines"—rather, it systematically transfers knowledge and decision-making ability through actual work experience.

Hanselman argues that the software industry should adopt this model. While AI automates routine tasks, veteran engineers should work closely with junior developers, patiently explaining "why we design it this way" and "what this code is meant to do," passing along their entire thought processes. This is the kind of structure he advocates.

This proposal matters because it reveals that AI adoption is about more than just "efficiency gains." When AI replaces lower-level tasks, the career development pathways that traditionally relied on such work as stepping stones become precarious. In other professions like medicine and law, apprenticeship periods are institutionalized, but the software industry has not yet developed comparable formal structures.

Going forward, the critical question is how companies and educational institutions will redesign junior development curricula. Cultivating both "the ability to leverage AI as a tool" and "the judgment to recognize AI's limitations" cannot be achieved through online learning content alone. Hanselman's proposal represents one starting point for broader industry-wide discussion about institutional design.

As AI advances, the contours of what humans must do become sharper. The junior engineer development challenge is simultaneously a technical issue and a cultural question: what kind of learning culture will organizations and the industry create? Engaging with this question may be essential to the software industry's sustainable future.