Know Your Qubits, Know Your Users: Personas for Quantum Software

AI summary

Overview

Research area: Quantum software engineering, with a focus on human-centered design and Model-Driven Engineering (MDE) — specifically, applying the interaction-design method of personas to characterize who actually uses quantum software.

Technical level: Beginner-Friendly to Intermediate. No quantum mechanics or programming background is required; the paper is qualitative and interview-based rather than mathematical.

Scope: A stakeholder analysis, based on an expert focus group at Dagstuhl Seminar 24512 (December 2024) and nine practitioner interviews at IEEE QCE 2025, that derives eleven aggregated personas for quantum software, each described by use case, interests, constraints, and desired level of abstraction.

What This Paper Is About

Quantum software is built by and for an unusually interdisciplinary mix of people — physicists, chemists, algorithm researchers, HPC engineers, business decision-makers — but knowledge about who these users are and what they need is fragmented across institutions and small specialist communities. Because a central design question in quantum software is how much hardware detail to expose versus hide, the authors argue that developers need a shared, consolidated picture of their users. The paper addresses this by conducting two qualitative studies and merging their results into eleven personas, each characterized along four dimensions (use cases, interests, constraints, and abstraction level), to serve as a common language for designing tailored quantum software, models, and domain-specific languages (DSLs).

Key Contributions

1. A focus-group study with quantum software experts. At Dagstuhl Seminar 24512 in December 2024, which brought together 24 experts for five days (three of them dedicated to working-group discussions), a group of eight to ten participants produced persona candidates categorized by short-, mid-, and long-term hardware maturity. The introduction reports this as five short-, six mid-, and five long-term personas (Section 4 later states five short-, seven mid-, and six long-term).

2. A practitioner interview study. At a workshop titled "Care and Maintenance of Quantum Software Stakeholders" at IEEE QCE 2025, the authors ran nine semi-structured one-on-one interviews lasting 15 to 40 minutes, with eleven demographic questionnaire responses collected; interviewees identified nine personas, weighted toward the short-term perspective.

3. A mapping between the two persona sets. Figure 3 relates Dagstuhl personas to interview personas using approximate-equivalence and subset relations, showing where the two independent sources converged and where they diverged.

4. Eleven aggregated personas with structured attributes. Table 2 gives each persona a use case (RQ2.1), interests (RQ2.2), constraints (RQ2.3), and abstraction level (RQ2.4), grouped by whether they are application oriented, application-and-hardware oriented, or hardware oriented.

Main Findings

• Eleven aggregated personas. The final set is: business user, early adopter, government contractor, simulation focused engineer, researchers (with physicist and chemist as subgroups), platform builder, quantum algorithm designer, HPC engineer, and embedded quantum developer.

• Three-tier grouping by orientation. Application-oriented personas (business user, early adopter, government contractor) want high abstraction and domain focus. Application-and-hardware-oriented personas (simulation focused engineer, researchers, physicist, chemist) sit in the middle. Hardware-oriented personas (platform builder, quantum algorithm designer, HPC engineer, embedded quantum developer) want low abstraction and access to hardware and system details.

• Convergence between the two studies. Business users, quantum algorithm designers, and HPC engineers were described roughly equivalently by both the Dagstuhl experts and the QCE interviewees, and were merged into single personas.

• Divergence between the two studies. Interviewees described physicists and chemists more generically than the Dagstuhl personas of quantum mechanics researchers and material scientists, so the authors use the broader physicist/chemist framing. Interviewees also contributed personas absent from the expert group: government contractor, simulation focused engineer, platform builder, and a broad "researcher" persona. Conversely, "early adopter" and "embedded quantum developer" were mentioned only by the expert group.

• Hardware maturity drives persona evolution. Short-term systems are described as having tens to hundreds of physical qubits, limiting execution to log-depth circuits and lacking interconnects for distributed QPUs. Mid-term systems are projected at thousands to hundreds of thousands of physical qubits with error-correcting codes that may not correct all errors but enable circuits beyond log depth with reasonable success rates, plus initial interconnects combining tens to hundreds of QPUs within this decade according to vendor roadmaps. The long-term goal is de-risked physics with reliable interconnects transmitting quantum information faithfully across QPUs.

• Business users are profit-driven across all horizons. Interviewees described them as spanning short-, mid-, and long-term use, with differences mainly in extent of application. Their use cases include acquiring new customers, protecting intellectual property (including security aspects of quantum computing), and driving research into industry applications. They want to abstract away syntax and implementation details, prefer natural language over rigorous formal languages, and rely on resource estimates — making them dependent on the accuracy and engineering effort behind those estimates.

• Researchers are fast-paced and resource-constrained. Their goals are showing quantum advantage and publishing findings, which requires a fast prototyping environment; time constraints on operations and personnel restrictions (favoring modular, low-granularity tasks over large multi-person work units) shape their tooling needs.

• Physicists and chemists sit at lower abstraction than business users. Physicists' interests include verifying theories, finding new phenomena, exploring condensed matter, and simulating problems such as in the aerospace domain. They span programming languages like Python down to quantum instructions and properties of quantum mechanics, and are constrained by manual transfer of physical details onto algorithms — highlighting the need for automation. Chemists target hard problems across physical, inorganic, organic, and biochemistry, constrained by physics and chemistry laws plus resource limitations that cause poor scalability of simulations.

• Hardware-oriented personas face integration and standardization constraints. Platform builders (of SDKs, APIs, abstraction frameworks) are constrained by modularity and replaceability of key toolchain components such as transpilers, hardware scalability, and a lack of software standardization. Quantum algorithm designers need access to backend-specific details and currently lack standardization. HPC engineers are constrained by communication time, including real-time connections. Government contractors require traceable, deterministic guarantees about software behavior and cannot rely on black-box libraries.

• Interdisciplinarity is empirically visible. The authors note that at the Munich Quantum Software Forum (MQSF), attendees with a physics background still outnumber those with a computer science background.

Methodology in Plain English

The authors borrow a technique from interaction design called Goal-Directed Design, in which designers first research real users and then distill them into personas — narrative descriptions of user types that act as a shared language between stakeholders, developers, and designers. Because the paper has no single concrete product in mind, the authors deliberately produce a persona-hypothesis rather than finished personas: an initial sketch of user kinds that future practitioners can refine into concrete personas for their own products.

Data came from two independent qualitative sources. First, a focus group of quantum software engineering experts at a 2024 Dagstuhl seminar, who were asked to identify stakeholder groups and project how they change as hardware matures, producing short-, mid-, and long-term variants. Second, semi-structured one-on-one interviews at IEEE QCE 2025, each with one interviewee, one interviewer, and one note-taker (three interviewers and three note-takers in total, all with academic backgrounds), guided by the research questions but open enough for free conversation. Interviewees were shown the Dagstuhl personas beforehand and invited to add new ones. Participants were screened via a demographic questionnaire confirming experience: of eleven respondents, five held a Master's degree and five a doctoral degree, and most reported expert-level experience in both quantum computing and software engineering. The two persona sets were then compared, matched by approximate equivalence or subset relations, and merged into eleven aggregated personas. The authors state that raw interview data cannot be shared for privacy reasons.

Why This Matters

Impact on research. MDE treats models as central artifacts created for a specific purpose and stakeholder, and DSLs are only as good as the understanding of who will use them. This paper supplies that missing user picture for quantum software, where the community has produced languages and frameworks (Qiskit and Cirq as widely adopted Python-embedded examples, plus special-purpose languages and the high-level Qrisp language) but has lacked a consolidated account of stakeholder needs. It also connects quantum software engineering to an established design methodology with a substantial literature.

Real-world applications:

• Guiding the design of DSLs and meta-models for quantum circuits, so that, for example, a physicist can express Hamiltonians in line with their mental model rather than in gate-level terms.
• Informing abstraction decisions in quantum SDKs and toolchains — when to expose error rates, topology, and qubit type for co-design, and when to hide them behind a domain interface.
• Supporting business decision-making, where personas need resource estimates and natural-language outputs rather than executable circuits, to judge how much of a process should run on quantum hardware.
• Structuring quantum-classical integration work in HPC settings, where engineers must interleave communication and coordinate computations across multiple QPUs.

Industry relevance. The persona set spans commercial, governmental, and technical roles, and the interviews were drawn from a conference attended by both academics and industry professionals. For companies building quantum platforms, the personas give a checklist of constraints — modularity, replaceability, standardization gaps, traceability, and security — that map directly onto product and interface decisions.

Future Directions

• Turning the persona-hypothesis into concrete personas. The authors explicitly frame their eleven personas as a starting point; future practitioners are expected to use them to plan initial interviews and then derive concrete personas for specific quantum software products.
• Adopting the full Goal-Directed Design process. The paper only uses the early research and modeling phases; later phases, in which personas drive interface design, remain to be applied to quantum software.
• Tracking personas as hardware matures. Because the short-, mid-, and long-term distinctions rest on projections (tens to hundreds of qubits now, thousands to hundreds of thousands mid-term, multi-QPU interconnects long-term), the persona set needs revisiting as those projections resolve.
• Broadening and validating the evidence base. The interview study rests on nine interviews and eleven demographic responses from a single venue, and raw data cannot be shared; wider sampling across venues and regions would test how well the aggregated personas generalize.

Target Audience

Quantum software engineers and architects deciding what to expose or hide in an SDK or language; MDE and DSL researchers who need to justify modeling choices against real stakeholder needs; interaction designers and usability researchers entering quantum computing; research group leaders and product managers who need a compact map of who their users are; and newcomers to quantum software who want a structured orientation to the field's many disciplines without wading into physics.

Authors’ abstract

The advancement of quantum hardware and the intricacies of quantum computing make well-designed quantum software increasingly necessary. Due to the interdisciplinarity of the field, it is crucial to understand the perspectives and specific needs of involved stakeholders, for example, to balance the desired level of abstraction with the exposition of (hardware)-specific details. In this work, we conduct a stakeholder-based analysis to identify personas of quantum software as a means of creating meaningful, user-tailored quantum software. We conducted an expert focus group at a Dagstuhl seminar in 2024 and qualitative interviews with practitioners at conference IEEE QCE in 2025, from which we derive eleven personas of potential users and stakeholders for quantum software. We discuss these personas regarding their use cases, interests, constraints and abstraction level.