# AI4Science — team work and attributable contribution

Reviewed: 2026-10-19. Technical overview only; private code, internal logs and scientific results are excluded.

## System inspected

Team repositories contain executable screening and molecular candidate workflows, explicit constraints, knowledge loading, generation/evaluation stages, containerized execution, scoring checks, and an optional LLM adapter with rule-based fallback. A working code path does not establish that an LLM improved scientific outcomes. No competition rank or biological efficacy is reported.

## Personal contribution

Nine LeadForge commits attributable to LumosNox-bamboo were inspected. They concern constraint/risk architecture, product and value communication figures, terminology revisions, competitive positioning and diagram rendering workflow. This supports a bounded architecture/communication contribution. It does not establish sole or primary implementation ownership of the candidate generation, molecular scoring or evaluation engines.

AI4Science is therefore presented as active team development (evidence C for the individual portfolio claim). Individual technical roles beyond this inspected record require confirmation. The competition archive is not counted as a second independent achievement.