Code of Leadership S2E8: Digital Teamlead or Can Developer Performance Be Measured by Code? (Filed under DevEx)
Code of Leadership S2E8: Digital Teamlead or Can Developer Performance Be Measured by Code? (Rubric #DevEx)
Are your developers working on 100%? And is it possible to answer this question by code at all - without tables, additional reports and subjective assessment of the head? And if there was a downturn in the work - to distinguish the underboot from a complex legashi, tech debt, unfamiliar technology or a month of heavy debugging?
6 augsta 17:00 in Moscow with me in live Will be Ivan Gel, founder of Dex, as part of the Code of Leadership podcast. We’ll talk about UpCore, a system the team calls “digital teammate.”
The idea is ambitious: analyze the code without additional reports from the developer, evaluate its complexity taking into account dozens of factors, compare the result with the grade and show the leader the reasons for the change in efficiency.
According to the UpCore team, the system can determine how long it would take for a developer of a certain level to work, see the share of preserved code, alterations and debugging, take into account the complexity of the architecture, legasi, bugs and technology of the project. The presentation stated accuracy at the level 85% relative to expert assessment, as well as an increase in efficiency on average 12% within three months after implementation.
This is where the most interesting conversation begins. Code is an important result of the work of an engineer, but not all of his work. Architectural solutions, team assistance, review, exploration, avoidable errors, and context complexity are poorly reduced to a single number. And the metric on which a grade, bonus, or layoff depend quickly becomes a goal for optimization.
Let's talk: What UpCore considers effective and how it normalizes different projects, stacks and task types Is it possible to automatically determine grade and labor intensity only by code; As in the current environment, when code is written with the help of AI, you can measure the effectiveness of a programmer. How to distinguish weak work from legasi, tech debt, complex core of the system and long-term debugging; On what data were checked the declared 85Percent accuracy and growth 12%; Does full transparency increase developer awareness or destroy team trust? How to protect such a system from cheating and the team itself from erroneous management conclusions; Where is the boundary between useful engineering telemetry and digital surveillance?
Come not to a product demonstration, but to an honest conversation about whether it is possible to make the work of the team more transparent without losing the context, trust and responsibility of the leader.
#AI4SDLC #Engineering #Management #Leadership #Metrics #DevTools