As part of a research collaboration with Standard Chartered Bank, we developed a Kubernetes scheduling extension that translates natural-language allocation hints into multi-objective scheduling decisions for cluster schedulers.
Score Extender Service
The architecture operates as a kube-scheduler extension that manages node selection while preserving native lifecycle management protocols. It applies tiered Pareto-dominance filtering to identify non-dominated node placements before final scoring.
This work resulted in multiple scientific publications in the enterprise AI domain, as well as a patent submission. The research output directly improved the bank’s standing in the Evident AI Banking Index. The details can be found at https://ieeexplore.ieee.org/document/11398069.