Inspiration
We wanted to explore quantum computing through a practical banking problem: choosing which customers to contact with a limited budget and capacity.
What it does
QYield learns from 300 customer records, ranks 100 prospects, and selects six from the top ten. It then estimates the chance that at least three will subscribe.
How we built it
We used Python, Qiskit, and scikit-learn, running everything locally. The pipeline combines quantum-derived prediction features, QAOA-based selection, and a six-qubit risk simulation.
Challenges we ran into
The small dataset made reliable probability estimates difficult. Enforcing constraints correctly and reducing circuit overhead also took careful testing.
Accomplishments that we're proud of
Our quantum optimizer matched the exact classical optimum. We also reduced its CX gate count by about 90% while preserving numerical agreement.
What we learned
More complex circuits do not automatically produce better results. Careful validation and clear assumptions matter as much as circuit design.
🔗 Report: https://altz211.github.io/qyield/report.html
