Inspiration
We wanted to make quantum optimization something you could try with a problem you already understood. If you can explain what needs to happen, what it costs, and which rules must hold, you should be able to explore how a quantum algorithm would approach the decision.
What it does
QuantumLoop turns a plain-English optimization request into a reviewed solver plan. It identifies what to optimize, which constraints must hold, and which inputs users can adjust. The plan records the proposed algorithm, the resources it needs, and how to check the result.
How we built it
We built the backend in Python and used Gemini with Pydantic schemas to extract objectives, constraints, and units from the user’s prompt. The planning stage uses Google Search grounding to research approaches, checks the draft against qubit and runtime limits, and saves the approved plan with its assumptions and verification criteria.
Challenges we ran into
The hardest part was deciding what to put on the quantum circuit. For our scheduling example, we designed classical preprocessing to rule out individually infeasible options. That leaves the quantum algorithm a smaller decision to make, including how to assign time slots without double-booking a shared resource.
Accomplishments that we're proud of
We turned a sample scheduling prompt into an approved plan for the Quantum Approximate Optimization Algorithm. We also ran a three-qubit entanglement circuit through Qiskit on Quantum Rings and generated a circuit diagram and a histogram of the measured outcomes.
What we learned
A measured result still needs to be decoded and checked against the original problem. For a small example, we can enumerate the feasible choices classically to find the exact best answer and measure how close the quantum result comes.
What's next for QuantumLoop
We’ll connect the approved plan to a simulated quantum solver and an independent verifier. Then we’ll build the interactive animation so users can change an input, recalculate, and see how the verified outcome changes.
