04/06/2026
The baseline for software engineering has shifted.
At MarsBased, we’ve just updated our performance review templates for all technical roles because the technical bar today is nothing like it was two years ago. AI tools have raised the floor, while fundamentals like clean code and ownership still matter, AI fluency is now a key evaluation criterion.
The challenge is measuring true AI proficiency without falling into the trap of blind copy-pasting, and continuing to avoid outdated metrics like commit counts or story points.
We’re choosing full transparency by sharing the exact framework we use to evaluate engineers, the tools we’ve standardized on (like Claude Code), and the human skills that AI will never replace.
Read the full breakdown on our blog: https://marsbased.com/blog/2026/06/04/how-we-evaluate-software-engineers-in-the-age-of-ai