Quantum Software Development Life-Cycle
Future Software Technologies
Semester programme:Master Applied IT
Project group members:Georgi Shterev
Project description
The main idea of the project is designing a quantum software engineering product by applying standart software engineering practices. The solution includes a pipeline for continuous integration and continuous deployment to a real quantum computer, which is the highlight of the project. The solution combines classical software engineering approach and a quantum algorithm to propose a Quantum SDLC which can be used universally when designing a quantum software application.
Context
The domain is Quantum Software Engineering. This project is used to enable people to think more practically when it comes to quantum development. Most researchers who study this topic focus only on algorithmic performance and do not create a solution ena-to-ena. My idea is to close this gap, and make researchers look from the practical side.
Results
The most important product outcome is a working hybrid quantum–classical QAOA prototype that implements the full workflow from domain modelling to encoding, solving, backend execution, experimentation, and evaluation. Its value is not only that it solves a Max-Cut version of an exam-scheduling problem, but that it does so as an engineered software system: modular, tested, reproducible, and backend-agnostic. The QUBO encoding is especially valuable because it acts as a stable interface between the problem and different solver types, allowing exact, gate-based QAOA, and potentially annealing-based solvers to share the same cost definition.
A second important product outcome is the engineering infrastructure around the prototype: the test suite, seeded experiment runner, CI pipeline, noisy-simulator rehearsal, and gated hardware execution. This gives the project practical value because it reduces accidental hardware spending, catches transition bugs before real-device execution, and makes results reproducible rather than dependent on a single run or notebook state.
The most important insight is that “more quantum depth” is not automatically better in the NISQ setting. The validation showed that QAOA quality improved at shallow/intermediate depth but degraded and became less stable at higher depth under a fixed optimiser budget. The hardware validation strengthened this insight: real IBM Quantum execution showed that useful probability mass could be concentrated on optimal solutions, but also confirmed that additional circuit depth is progressively consumed by noise.
In terms of Technology Readiness Level, the work is best positioned around TRL 5. The prototype has moved beyond a basic proof of concept because it is implemented, tested, and validated across noiseless simulation, noisy simulation, and a real IBM Quantum backend. However, it is not yet TRL 6 or higher because the main 14-node study is still simulator-based, the hardware validation is limited to a smaller instance, and the exam-scheduling problem remains a proxy rather than a real operational timetabling system. Therefore, the project’s value lies in demonstrating a validated QSE approach that is ready for further scaling and realistic case-study validation.