Andriy Miranskyy

Dongchan Kim, Hamidreza Khoramrokh, Lei Zhang et al.

Flaky tests, tests that pass or fail nondeterministically without changes to code or environment, pose a serious threat to software reliability. While classical software engineering has developed a rich body of techniques to study flakiness, corresponding evidence for quantum software remains limited. Prior work relies mainly on static analysis or small sets of manually reported incidents, leaving open questions about their prevalence, characteristics, and detectability. This paper presents the first large-scale dynamic characterization of flaky tests in quantum software, focusing on the Qiskit Terra core library. We executed the Qiskit Terra test suite 10,000 times across 23 releases in controlled environments. For each release, we measured test-outcome variability, identified flaky tests, estimated empirical failure probabilities, analyzed recurrence across versions, used Wilson confidence intervals to quantify rerun budgets for reliable detection, and mapped flaky tests to Terra subcomponents. Across 27,026 fully qualified test identifiers, we identified 62 unique flaky tests. Although overall flakiness rates were low (0-0.17%), recurrence was substantial: 52 of 62 flaky tests (83.87%) reappeared in multiple releases, while only 10 tests (16.13%) were confined to a single release. Empirical failure probabilities spanned several orders of magnitude, with a median of $9 \times 10^{-4}$ and 34 tests (54.84%) at or below $10^{-3}$, implying that thousands to tens of thousands of executions may be required for confident detection. These results show that quantum test flakiness is rare but difficult to detect under typical continuous integration budgets. To support future research, we release a public dataset of per-test execution outcomes.

Andriy Miranskyy

Quantum computing is increasingly practiced through programming, yet most educational offerings emphasize algorithmic or framework-level use rather than software engineering concerns such as testing, abstraction, tooling, and lifecycle management. This paper reports on the design and first offering of a cross-listed undergraduate--graduate course that frames quantum computing through a software engineering lens, focusing on early-stage competence relevant to software engineering practice. The course integrates foundational quantum concepts with software engineering perspectives, emphasizing executable artifacts, empirical reasoning, and trade-offs arising from probabilistic behaviour, noise, and evolving toolchains. Evidence is drawn from instructor observations, supplemented by anonymous student feedback, a background survey, and inspection of student work. Despite minimal prior exposure to quantum computing, students were able to engage productively with quantum software engineering topics once a foundational understanding of quantum information and quantum algorithms, expressed through executable artifacts, was established. This experience report contributes a modular course design, a scalable assessment model for mixed academic levels, and transferable lessons for software engineering educators developing quantum computing curricula.

Lei Zhang, Sean Howard, Tom Montpool et al.

There has been a massive explosion of data generated by customers and retained by companies in the last decade. However, there is a significant mismatch between the increasing volume of data and the lack of automation methods and tools. The lack of best practices in data science programming may lead to software quality degradation, release schedule slippage, and budget overruns. To mitigate these concerns, we would like to bring software engineering best practices into data science. Specifically, we focus on automated data validation in the data preparation phase of the software development life cycle. This paper studies a real-world industrial case and applies software engineering best practices to develop an automated test harness called RESTORE. We release RESTORE as an open-source R package. Our experience report, done on the geodemographic data, shows that RESTORE enables efficient and effective detection of errors injected during the data preparation phase. RESTORE also significantly reduced the cost of testing. We hope that the community benefits from the open-source project and the practical advice based on our experience.

Andriy Miranskyy, Adam Sorrenti, Viral Thakar

The effectiveness of training neural networks directly impacts computational costs, resource allocation, and model development timelines in machine learning applications. An optimizer's ability to train the model adequately (in terms of trained model performance) depends on the model's initial weights. Model weight initialization schemes use pseudorandom number generators (PRNGs) as a source of randomness. We investigate whether substituting PRNGs for low-discrepancy quasirandom number generators (QRNGs) -- namely Sobol' sequences -- as a source of randomness for initializers can improve model performance. We examine Multi-Layer Perceptrons (MLP), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Transformer architectures trained on MNIST, CIFAR-10, and IMDB datasets using SGD and Adam optimizers. Our analysis uses ten initialization schemes: Glorot, He, Lecun (both Uniform and Normal); Orthogonal, Random Normal, Truncated Normal, and Random Uniform. Models with weights set using PRNG- and QRNG-based initializers are compared pairwise for each combination of dataset, architecture, optimizer, and initialization scheme. Our findings indicate that QRNG-based neural network initializers either reach a higher accuracy or achieve the same accuracy more quickly than PRNG-based initializers in 60% of the 120 experiments conducted. Thus, using QRNG-based initializers instead of PRNG-based initializers can speed up and improve model training.

Montgomery Gole, Williams-Paul Nwadiugwu, Andriy Miranskyy

Sarcasm is a form of irony that requires readers or listeners to interpret its intended meaning by considering context and social cues. Machine learning classification models have long had difficulty detecting sarcasm due to its social complexity and contradictory nature. This paper explores the applications of the Generative Pretrained Transformer (GPT) models, including GPT-3, InstructGPT, GPT-3.5, and GPT-4, in detecting sarcasm in natural language. It tests fine-tuned and zero-shot models of different sizes and releases. The GPT models were tested on the political and balanced (pol-bal) portion of the popular Self-Annotated Reddit Corpus (SARC 2.0) sarcasm dataset. In the fine-tuning case, the largest fine-tuned GPT-3 model achieves accuracy and $F_1$-score of 0.81, outperforming prior models. In the zero-shot case, one of GPT-4 models yields an accuracy of 0.70 and $F_1$-score of 0.75. Other models score lower. Additionally, a model's performance may improve or deteriorate with each release, highlighting the need to reassess performance after each release.

Mohammad Saiful Islam, Mohamed Sami Rakha, William Pourmajidi et al.

As Large-Scale Cloud Systems (LCS) become increasingly complex, effective anomaly detection is critical for ensuring system reliability and performance. However, there is a shortage of large-scale, real-world datasets available for benchmarking anomaly detection methods. To address this gap, we introduce a new high-dimensional dataset from IBM Cloud, collected over 4.5 months from the IBM Cloud Console. This dataset comprises 39,365 rows and 117,448 columns of telemetry data. Additionally, we demonstrate the application of machine learning models for anomaly detection and discuss the key challenges faced in this process. This study and the accompanying dataset provide a resource for researchers and practitioners in cloud system monitoring. It facilitates more efficient testing of anomaly detection methods in real-world data, helping to advance the development of robust solutions to maintain the health and performance of large-scale cloud infrastructures.

Janakan Sivaloganathan, Ainaz Jamshidi, Andriy Miranskyy et al.

Flaky tests, which pass or fail inconsistently without code changes, are a major challenge in software engineering in general and in quantum software engineering in particular due to their complexity and probabilistic nature, leading to hidden issues and wasted developer effort. We aim to create an automated framework to detect flaky tests in quantum software and an extended dataset of quantum flaky tests, overcoming the limitations of manual methods. Building on prior manual analysis of 14 quantum software repositories, we expanded the dataset and automated flaky test detection using transformers and cosine similarity. We conducted experiments with Large Language Models (LLMs) from the OpenAI GPT and Meta LLaMA families to assess their ability to detect and classify flaky tests from code and issue descriptions. Embedding transformers proved effective: we identified 25 new flaky tests, expanding the dataset by 54%. Top LLMs achieved an F1-score of 0.8871 for flakiness detection but only 0.5839 for root cause identification. We introduced an automated flaky test detection framework using machine learning, showing promising results but highlighting the need for improved root cause detection and classification in large quantum codebases. Future work will focus on improving detection techniques and developing automatic flaky test fixes.

Kiana Kheiri, Aamna Aamir, Andriy Miranskyy et al.

Quantum circuits must be error-resilient, yet LLMs like Granite-20B-Code and StarCoder often output flawed Qiskit code. We fine-tuned the Qwen2.5-Coder-32B model with two RL methods, Group Relative Policy Optimization (GRPO) and Odds-Ratio Preference Optimization (ORPO), using a richly annotated synthetic dataset. On the Qiskit HumanEval benchmark, ORPO reaches 56.29% Pass@1 ($\approx+10$ pp over Granite-8B-QK) and GRPO hits 49%, both beating all general-purpose baselines; on the original HumanEval they score 65.90% and 63.00%. GRPO performs well on basic tasks (44/78) and excels on intermediate ones (41/68), but neither GRPO nor ORPO solves any of the five advanced tasks, highlighting clear gains yet room for progress in AI-assisted quantum programming.

Montgomery Gole, Andriy Miranskyy

Sarcasm detection is challenging for both humans and machines. This work explores how model characteristics impact sarcasm detection in OpenAI's GPT, and Meta's Llama-2 models, given their strong natural language understanding, and popularity. We evaluate fine-tuned and zero-shot models across various sizes, releases, and hyperparameters. Experiments were conducted on the political and balanced (pol-bal) portion of the popular Self-Annotated Reddit Corpus (SARC2.0) sarcasm dataset. Fine-tuned performance improves monotonically with model size within a model family, while hyperparameter tuning also impacts performance. In the fine-tuning scenario, full precision Llama-2-13b achieves state-of-the-art accuracy and $F_1$-score, both measured at 0.83, comparable to average human performance. In the zero-shot setting, one GPT-4 model achieves competitive performance to prior attempts, yielding an accuracy of 0.70 and an $F_1$-score of 0.75. Furthermore, a model's performance may increase or decline with each release, highlighting the need to reassess performance after each release.

Mushahid Khan, Jean Paul Latyr Faye, Udson C. Mendes et al.

Machine learning (ML) classification tasks can be carried out on a quantum computer (QC) using Probabilistic Quantum Memory (PQM) and its extension, Parameteric PQM (P-PQM) by calculating the Hamming distance between an input pattern and a database of $r$ patterns containing $z$ features with $a$ distinct attributes. For accurate computations, the feature must be encoded using one-hot encoding, which is memory-intensive for multi-attribute datasets with $a>2$. We can easily represent multi-attribute data more compactly on a classical computer by replacing one-hot encoding with label encoding. However, replacing these encoding schemes on a QC is not straightforward as PQM and P-PQM operate at the quantum bit level. We present an enhanced P-PQM, called EP-PQM, that allows label encoding of data stored in a PQM data structure and reduces the circuit depth of the data storage and retrieval procedures. We show implementations for an ideal QC and a noisy intermediate-scale quantum (NISQ) device. Our complexity analysis shows that the EP-PQM approach requires $O\left(z \log_2(a)\right)$ qubits as opposed to $O(za)$ qubits for P-PQM. EP-PQM also requires fewer gates, reducing gate count from $O\left(rza\right)$ to $O\left(rz\log_2(a)\right)$. For five datasets, we demonstrate that training an ML classification model using EP-PQM requires 48% to 77% fewer qubits than P-PQM for datasets with $a>2$. EP-PQM reduces circuit depth in the range of 60% to 96%, depending on the dataset. The depth decreases further with a decomposed circuit, ranging between 94% and 99%. EP-PQM requires less space; thus, it can train on and classify larger datasets than previous PQM implementations on NISQ devices. Furthermore, reducing the number of gates speeds up the classification and reduces the noise associated with deep quantum circuits. Thus, EP-PQM brings us closer to scalable ML on a NISQ device.

Lei Zhang, Andriy Miranskyy, Walid Rjaibi et al.

The software engineering community is facing challenges from quantum computers (QCs). In the era of quantum computing, Shor's algorithm running on QCs can break asymmetric encryption algorithms that classical computers practically cannot. Though the exact date when QCs will become "dangerous" for practical problems is unknown, the consensus is that this future is near. Thus, the software engineering community needs to start making software ready for quantum attacks and ensure quantum safety proactively. We argue that the problem of evolving existing software to quantum-safe software is very similar to the Y2K bug. Thus, we leverage some best practices from the Y2K bug and propose our roadmap, called 7E, which gives developers a structured way to prepare for quantum attacks. It is intended to help developers start planning for the creation of new software and the evolution of cryptography in existing software. In this paper, we use a case study to validate the viability of 7E. Our software under study is the IBM Db2 database system. We upgrade the current cryptographic schemes to post-quantum cryptographic ones (using Kyber and Dilithium schemes) and report our findings and lessons learned. We show that the 7E roadmap effectively plans the evolution of existing software security features towards quantum safety, but it does require minor revisions. We incorporate our experience with IBM Db2 into the revised 7E roadmap. The U.S. Department of Commerce's National Institute of Standards and Technology is finalizing the post-quantum cryptographic standard. The software engineering community needs to start getting prepared for the quantum advantage era. We hope that our experiential study with IBM Db2 and the 7E roadmap will help the community prepare existing software for quantum attacks in a structured manner.

Andriy Miranskyy, Lei Zhang, Javad Doliskani

Quantum computers are becoming more mainstream. As more programmers are starting to look at writing quantum programs, they need to test and debug their code. In this paper, we discuss various use-cases for quantum computers, either standalone or as part of a System of Systems. Based on these use-cases, we discuss some testing and debugging tactics that one can leverage to ensure the quality of the quantum software. We also highlight quantum-computer-specific issues and list novel techniques that are needed to address these issues. The practitioners can readily apply some of these tactics to their process of writing quantum programs, while researchers can learn about opportunities for future work.

William Pourmajidi, Lei Zhang, John Steinbacher et al.

Service Level Agreements (SLA) are employed to ensure the performance of Cloud solutions. When a component fails, the importance of logs increases significantly. All departments may turn to logs to determine the cause of the issue and find the party at fault. The party at fault may be motivated to tamper with the logs to hide their role. We argue that the critical nature of Cloud logs calls for immutability and verification mechanism without the presence of a single trusted party. This paper proposes such a mechanism by describing a blockchain-based log storage system, called Logchain, which can be integrated with existing private and public blockchain solutions. Logchain uses the immutability feature of blockchain to provide a tamper-resistance platform for log storage. Additionally, we propose a hierarchical structure to address blockchains' scalability issues. To validate the mechanism, we integrate Logchain into Ethereum and IBM Blockchain. We show that the solution is scalable and perform the analysis of the cost of ownership to help a reader select an implementation that would address their needs. The Logchain's scalability improvement on a blockchain is achieved without any alteration of blockchains' fundamental architecture. As shown in this work, it can function on private and public blockchains and, therefore, can be a suitable alternative for organizations that need a secure, immutable log storage platform.

Mohammad Saiful Islam, Andriy Miranskyy

Cloud platforms, under the hood, consist of a complex inter-connected stack of hardware and software components. Each of these components can fail which may lead to an outage. Our goal is to improve the quality of Cloud services through early detection of such failures by analyzing resource utilization metrics. We tested Gated-Recurrent-Unit-based autoencoder with a likelihood function to detect anomalies in various multi-dimensional time series and achieved high performance.

Andriy Miranskyy, Lei Zhang, Javad Doliskani

Quantum computers are becoming more mainstream. As more programmers are starting to look at writing quantum programs, they face an inevitable task of debugging their code. How should the programs for quantum computers be debugged? In this paper, we discuss existing debugging tactics, used in developing programs for classic computers, and show which ones can be readily adopted. We also highlight quantum-computer-specific debugging issues and list novel techniques that are needed to address these issues. The practitioners can readily apply some of these tactics to their process of writing quantum programs, while researchers can learn about opportunities for future work.

William Pourmajidi, Andriy Miranskyy, John Steinbacher et al.

The stability and performance of Cloud platforms are essential as they directly impact customers' satisfaction. Cloud service providers use Cloud monitoring tools to ensure that rendered services match the quality of service requirements indicated in established contracts such as service-level agreements. Given the enormous number of resources that need to be monitored, highly scalable and capable monitoring tools are designed and implemented by Cloud service providers such as Amazon, Google, IBM, and Microsoft. Cloud monitoring tools monitor millions of virtual and physical resources and continuously generate logs for each one of them. Considering that logs magnify any technical issue, they can be used for disaster detection, prevention, and recovery. However, logs are useless if they are not assessed and analyzed promptly. Thus, we argue that the scale of Cloud-generated logs makes it impossible for DevOps teams to analyze them effectively. This implies that one needs to automate the process of monitoring and analysis (e.g., using machine learning and artificial intelligence). If the automation will witness an anomaly in the logs --- it will alert DevOps staff. The automatic anomaly detectors require a reliable and scalable platform for gathering, filtering, and transforming the logs, executing the detector models, and sending out the alerts to the DevOps staff. In this work, we report on implementing a prototype of such a platform based on the 7-layered architecture pattern, which leverages micro-service principles to distribute tasks among highly scalable, resources-efficient modules. The modules interact with each other via an instance of the Publish-Subscribe architectural pattern. The platform is deployed on the IBM Cloud service infrastructure and is used to detect anomalies in logs emitted by the IBM Cloud services, hence the dogfooding.

William Pourmajidi, John Steinbacher, Tony Erwin et al.

Cloud services are becoming increasingly popular: 60\% of information technology spending in 2016 was Cloud-based, and the size of the public Cloud service market will reach \$236B by 2020. To ensure reliable operation of the Cloud services, one must monitor their health. While a number of research challenges in the area of Cloud monitoring have been solved, problems are remaining. This prompted us to highlight three areas, which cause problems to practitioners and require further research. These three areas are as follows: A) defining health states of Cloud systems, B) creating unified monitoring environments, and C) establishing high availability strategies. In this paper we provide details of these areas and suggest a number of potential solutions to the challenges. We also show that Cloud monitoring presents exciting opportunities for novel research and practice.

William Pourmajidi, Andriy Miranskyy

During the normal operation of a Cloud solution, no one usually pays attention to the logs except technical department, which may periodically check them to ensure that the performance of the platform conforms to the Service Level Agreements. However, the moment the status of a component changes from acceptable to unacceptable, or a customer complains about accessibility or performance of a platform, the importance of logs increases significantly. Depending on the scope of the issue, all departments, including management, customer support, and even the actual customer, may turn to logs to find out what has happened, how it has happened, and who is responsible for the issue. The party at fault may be motivated to tamper the logs to hide their fault. Given the number of logs that are generated by the Cloud solutions, there are many tampering possibilities. While tamper detection solution can be used to detect any changes in the logs, we argue that critical nature of logs calls for immutability. In this work, we propose a blockchain-based log system, called Logchain, that collects the logs from different providers and avoids log tampering by sealing the logs cryptographically and adding them to a hierarchical ledger, hence, providing an immutable platform for log storage.

Andriy V. Miranskyy, Zainab Al-zanbouri, David Godwin et al.

Context: Information Technology consumes up to 10\% of the world's electricity generation, contributing to CO2 emissions and high energy costs. Data centers, particularly databases, use up to 23% of this energy. Therefore, building an energy-efficient (green) database engine could reduce energy consumption and CO2 emissions. Goal: To understand the factors driving databases' energy consumption and execution time throughout their evolution. Method: We conducted an empirical case study of energy consumption by two MySQL database engines, InnoDB and MyISAM, across 40 releases. We examined the relationships of four software metrics to energy consumption and execution time to determine which metrics reflect the greenness and performance of a database. Results: Our analysis shows that database engines' energy consumption and execution time increase as databases evolve. Moreover, the Lines of Code metric is correlated moderately to strongly with energy consumption and execution time in 88% of cases. Conclusions: Our findings provide insights to both practitioners and researchers. Database administrators may use them to select a fast, green release of the MySQL database engine. MySQL database-engine developers may use the software metric to assess products' greenness and performance. Researchers may use our findings to further develop new hypotheses or build models to predict greenness and performance of databases.

Sokratis Tsakiltsidis, Andriy Miranskyy, Elie Mazzawi

Context: Software performance is a critical non-functional requirement, appearing in many fields such as mission critical applications, financial, and real time systems. In this work we focused on early detection of performance bugs; our software under study was a real time system used in the advertisement/marketing domain. Goal: Find a simple and easy to implement solution, predicting performance bugs. Method: We built several models using four machine learning methods, commonly used for defect prediction: C4.5 Decision Trees, Naïve Bayes, Bayesian Networks, and Logistic Regression. Results: Our empirical results show that a C4.5 model, using lines of code changed, file's age and size as explanatory variables, can be used to predict performance bugs (recall=0.73, accuracy=0.85, and precision=0.96). We show that reducing the number of changes delivered on a commit, can decrease the chance of performance bug injection. Conclusions: We believe that our approach can help practitioners to eliminate performance bugs early in the development cycle. Our results are also of interest to theoreticians, establishing a link between functional bugs and (non-functional) performance bugs, and explicitly showing that attributes used for prediction of functional bugs can be used for prediction of performance bugs.

Mujahid Sultan, Andriy Miranskyy

Context: Enterprise Architecture (EA) is a discipline which has evolved to structure the business and its alignment with the IT systems. One of the popular enterprise architecture frameworks is Zachman framework (ZF). This framework focuses on describing the enterprise from six viewpoint perspectives of the stakeholders. These six perspectives are based on English language interrogatives 'what', 'where', 'who', 'when', 'why', and 'how' (thus the term W5H Journalists and police investigators use the W5H to describe an event. However, EA is not an event, creation and evolution of EA challenging. Moreover, the ordering of viewpoints is not defined in the existing EA frameworks, making data capturing process difficult. Our goals are to 1) assess if W5H is sufficient to describe modern EA and 2) explore the ordering and precedence among the viewpoint concerns. Method: we achieve our goals by bringing tools from the Linguistics, focusing on a full set of English Language interrogatives to describe viewpoint concerns and the inter-relationships and dependencies among these. Application of these tools is validated using pedagogical EA examples. Results: 1) We show that addition of the seventh interrogative 'which' to the W5H set (we denote this extended set as W6H) yields extra and necessary information enabling creation of holistic EA. 2) We discover that particular ordering of the interrogatives, established by linguists (based on semantic and lexical analysis of English language interrogatives), define starting points and the order in which viewpoints should be arranged for creating complete EA. 3) We prove that adopting W6H enables creation of EA for iterative and agile SDLCs, e.g. Scrum. Conclusions: We believe that our findings complete creation of EA using ZF by practitioners, and provide theoreticians with tools needed to improve other EA frameworks, e.g., TOGAF and DoDAF.

Mujahid Sultan, Andriy Miranskyy

Requirements elicitation and requirements analysis are important practices of Requirements Engineering. Elicitation techniques, such as interviews and questionnaires, rely on formulating interrogative questions and asking these in a proper order to maximize the accuracy of the information being gathered. Information gathered during requirements elicitation then has to be interpreted, analyzed, and validated. Requirements analysis involves analyzing the problem and solutions spaces. In this paper, we describe a method to formulate interrogative questions for effective requirements elicitation based on the lexical and semantic principles of the English language interrogatives, and propose a pattern to organize stakeholder viewpoint concerns for better requirements analysis. This helps requirements engineer thoroughly describe problem and solutions spaces. Most of the previous requirements elicitation studies included six out of the seven English language interrogatives 'what', 'where', 'when', 'who', 'why', and 'how' (denoted by W5H) and did not propose any order in the interrogatives. We show that extending the set of six interrogatives with 'which' (denoted by W6H) improves the generation and formulation of questions for requirements elicitation and facilitates better requirements analysis via arranging stakeholder views. We discuss the interdependencies among interrogatives (for requirements engineer to consider while eliciting the requirements) and suggest an order for the set of W6H interrogatives. The proposed W6H-based reusable pattern also aids requirements engineer in organizing viewpoint concerns of stakeholders, making this pattern an effective tool for requirements analysis.