A Junior Without Easy Tasks. How Do You Enter Tech in the AI Era?

The co-hosts begin with the familiar ladder from university and an internship to small tasks and growing independence. When agents absorb simple fixes, a business has fewer reasons to pay for that traditional learning stage. Large organizations may still develop strong interns, but a program for programming-contest participants does not represent every beginner. At industry level, a collective problem emerges: everyone needs future experienced engineers, while an individual employer may prefer to hire an established specialist and equip them with better tools. Analogies with game theory and the tragedy of the commons expose the gap between short-term incentives and shared interests.

The conversation separates two meanings of entering tech: starting to create products and getting paid for doing so. For a self-directed learner, AI provides explanations and makes a complete project lifecycle accessible. But a beginner without domain knowledge competes with both an experienced engineer and a business specialist automating their own work. Professional relationships, interest in users, and problem understanding therefore remain valuable. Engineering fundamentals must be learned alongside the tools; working at a higher level of abstraction does not remove the need to reason about a solution and verify it.

Aleksey proposes a completed project and a discussion of how it was built as evidence of readiness. Agent instructions, automated checks, CI/CD, model evaluation, and the response to errors reveal independence. The point is not a technology checklist but the ability to explain decisions and show how the result was verified. The emerging profile combines interest in a business problem, systematic learning, and responsibility. This is not presented as a uniform requirement at every employer: companies differ in how they assess beginners and how much they are willing to invest in their development.

The final part focuses on learning: a book’s table of contents reveals unknown areas, while questions and examples help investigate them with AI. Receiving an artifact still does not prove understanding. Alexander proposes a deliberately designed learning cycle: the beginner explains the task and acceptance criteria, defends the agent’s solution to a mentor, and understands release and rollback. Mastering the company’s own process may be enough, rather than learning every methodology at once. The co-hosts consider publicly released development workflows an additional environment for independent practice, while retaining the value of mentorship and feedback.