Researchers have published a constraint-programming framework that tries to make workforce scheduling less of a legal liability and an optimization nightmare.
CP-WSP is a declarative framework built on CP-SAT — a type of solver that treats hard rules as mathematically unbreakable — and it handles 14 hard constraints (think: mandatory break placement, cross-midnight shift patterns, regulatory coverage floors) alongside 15 softer objectives like workload equity and schedule stability. Operators configure the whole thing through a JSON file, with no code changes required when switching between scheduling setups. The team evaluated it against the INRC-II benchmark suite, a standard reference dataset for nurse rostering, across 36 synthetic configurations at both hourly and shift-level granularity.
The gap it targets is real. Most published CP formulations for workforce scheduling cap out at six to twelve constraints and treat shifts as indivisible blocks — fine for academic papers, less fine for a hospital running 24-hour rotations with acuity-weighted patient loads. By operating at sub-shift granularity (down to 30-minute intervals), CP-WSP can match staffing to demand in ways that shift-level models simply cannot express. The zero-regulatory-violations-by-construction claim is the kind of thing that gets the attention of compliance teams, if the benchmarks hold up in production environments.
Scheduling optimization is a crowded research area, and the distance between a promising benchmark result and a deployed enterprise tool is significant — most prior frameworks have died somewhere in that gap.