
The Junior Engineer After Code
Growing engineers when agents do the implementation

Growing engineers when agents do the implementation
Growing engineers when agents do the implementation
Ownership becomes visible after the first version
Output, learning, and ownership cannot be collapsed
Overall and junior estimates are not interchangeable
Job seniority does not define expertise in a new domain
A cheap local step still meets the whole system
Task speed and acquired skill can diverge
It no longer emerges automatically from the task queue
Agents may absorb safe entry-level work
Preserve effort where causal understanding is formed
The agent executes; the human owns criteria and consequences
The signal is in the model, falsification, and transfer
The generator must not accept its own result alone
Working hypothesis: level follows the autonomy radius
Fragment → task → change → episode → component
Ownership ends with an observed outcome
Trade-offs, operations, and incidents are part of the job
Direction matters more than one successful project
Expose the work loop rather than typing speed
Plan, history, diff, checks, and a changed constraint
The same artifact can carry opposite signals
A role without apprenticeship exports cost to seniors
Its owners, cadences, and measures must be explicit
Mentor, manager, and platform are not interchangeable
Value today and independence tomorrow
The agent accelerates work. The team grows the engineer.
The Junior Engineer After Code
The longread is linked on slide one; new materials are in Knizhny kub
Alexander Polomodov, Technical Director & Fellow, T-Technologies
@book_cube