Case 01
Reliable Workforce Scheduling Under Uncertainty
How can interdependent staffing rules become a decision-support workflow that responds honestly when capacity is insufficient?
- Coverage
- Qualifications
- Leave
- Rest rules
- Soft constraints
- Capacity limits
What I built directly
I personally developed the nurse MILP/Python model and the patient-services staff scheduling model. I translated coverage, qualification, leave and rest requirements into mathematical constraints, determined data-informed objective weights from the supplied data, and supported the team's data preparation, benchmarking and analysis.
How I organised the delivery
Alongside my own modelling work, I coordinated task allocation across the seven-person team, consolidated the report and presentation, and presented the final work. Some assigned tasks were delayed or left incomplete, so keeping ownership visible and bringing separate contributions into one coherent deliverable became a real part of the project.
Under matched leave conditions, the report's model comparison found 42.31% lower night-shift deviation, 51.92% lower weekend-shift deviation and 57.69% lower monthly average overtime. The comparison tests whether the model improves workload balance and overtime performance against the historical scheduling baseline.
Operational records + rules
Six months of company-provided shift data, qualification information and scheduling rules.
Two scheduling systems
Nurse MILP/Python and patient-services staff scheduling developed directly by me.
Feasibility under stress
Constraint checks, small-instance exhaustive comparison and high-severity scenarios tested whether rules held and when capacity was insufficient.
Decision-support prototype
Built and tested as a capstone output, with capacity limits made explicit instead of hidden behind a forced schedule.
When delivery slipped
- Challenge
- Assigned tasks were sometimes delayed or not completed.
- Response
- Coordinated task ownership and pulled the report and presentation into one final output.
- Outcome
- The team submitted and presented an integrated capstone deliverable.
- What I learned
- Unclear ownership becomes an analytical, integration and deadline risk—not only a people issue.