The recent advancements of Small Language Models (SLMs) have opened new possibilities for efficient code generation. SLMs offer lightweight and cost-effective alternatives to Large Language Models (LLMs), making them attractive for use in resource-constrained environments. However, empirical understanding of SLMs, particularly their capabilities, limitations, and performance trade-offs in code generation remains limited. This study presents a comprehensive empirical evaluation of 20 open-source SLMs ranging from 0.4B to 10B parameters on five diverse code-related benchmarks (HumanEval, MBPP, Mercury, HumanEvalPack, and CodeXGLUE). The models are assessed along three dimensions: i) functional correctness of generated code, ii) computational efficiency and iii) performance across multiple programming languages. The findings of this study reveal that several compact SLMs achieve competitive results while maintaining a balance between performance and efficiency, making them viable for deployment in resource-constrained environments. However, achieving further improvements in accuracy requires switching to larger models. These models generally outperform their smaller counterparts, but they require much more computational power. We observe that for 10% performance improvements, models can require nearly a 4x increase in VRAM consumption, highlighting a trade-off between effectiveness and scalability. Besides, the multilingual performance analysis reveals that SLMs tend to perform better in languages such as Python, Java, and PHP, while exhibiting relatively weaker performance in Go, C++, and Ruby. However, statistical analysis suggests these differences are not significant, indicating a generalizability of SLMs across programming languages. Based on the findings, this work provides insights into the design and selection of SLMs for real-world code generation tasks.
Mahade Hasan, Muhammad Waseem, Kai-Kristian Kemell et al.· Journal of Systems and Softw...· 16 citations
The use of Large Language Models (LLMs) for autonomous code generation is gaining attention in emerging technologies. As LLM capabilities expand, they offer new possibilities such as code refactoring, security enhancements, and legacy application upgrades. Many outdated web applications pose security and reliability challenges, yet companies continue using them due to the complexity and cost of upgrades. To address this, we propose an LLM-based multi-agent system that autonomously upgrades legacy web applications to the latest versions. The system distributes tasks across multiple phases, updating all relevant files. To evaluate its effectiveness, we employed Zero-Shot Learning (ZSL) and One-Shot Learning (OSL) prompts, applying identical instructions in both cases. The evaluation involved updating view files and measuring the number and types of errors in the output. For complex tasks, we counted the successfully met requirements. The experiments compared the proposed system with standalone LLM execution, repeated multiple times to account for stochastic behavior. Results indicate that our system maintains context across tasks and agents, improving solution quality over the base model in some cases. This study provides a foundation for future model implementations in legacy code updates. Additionally, findings highlight LLMs' ability to update small outdated files with high precision, even with basic prompts. The source code is publicly available on GitHub: https://github.com/alasalm1/Multi-agent-pipeline.
Valtteri Ala-Salmi, Z. Rasheed, Malik Abdul Sami et al.· International Conference on...· 4 citations
Vibe Coding (VC) is a form of software development assisted by generative AI, in which developers describe the intended functionality or logic via natural language prompts, and the AI system generates the corresponding source code. VC can be leveraged for rapid prototyping or developing the Minimum Viable Products (MVPs); however, it may introduce several risks throughout the software development life cycle. Based on our experience from several internally developed MVPs and a review of recent industry reports, this article analyzes the flow-debt tradeoffs associated with VC. The flow-debt trade-off arises when the seamless code generation occurs, leading to the accumulation of technical debt through architectural inconsistencies, security vulnerabilities, and increased maintenance overhead. These issues originate from process-level weaknesses, biases in model training data, a lack of explicit design rationale, and a tendency to prioritize quick code generation over human-driven iterative development. Based on our experiences, we identify and explain how current model, platform, and hardware limitations contribute to these issues, and propose countermeasures to address them, informing research and practice towards more sustainable VC approaches.
Muhammad Waseem, Aakash Ahmad, Kai-Kristian Kemell et al.· arXiv.org· 4 citations
Vibe coding is a software development practice in which developers state intent in natural language and large language models generate code. It is often framed as one-shot prompting, but the evidence describes an intent-driven, iterative workflow whose outcomes depend on how generated code is evaluated and governed. Knowledge of how vibe coding is defined, practiced, and governed is scattered across academic and practitioner sources, and, to our knowledge, existing reviews have not yet integrated both evidence streams. We conducted a multivocal literature review of peer-reviewed and grey literature following established guidelines. Searches spanned 2022 to October 2025. After screening, credibility assessment, and snowballing, 47 sources were retained (28 peer-reviewed and 19 grey) and analyzed through descriptive mapping and thematic synthesis across eight research questions. Vibe coding is consistently described as an iterative generation-evaluation-revision loop rather than a one-shot activity, and developer work shifts from writing code towards specification, supervision, and validation. Short-term productivity and time-to-prototype gains are reported in 21 of 47 sources (45%), while evidence on maintainability, long-term quality, and safeguard effectiveness remains limited. Evidence is strongest for prototyping and user-interface work and weakest for production, data-intensive, and safety-critical use, and tool visibility does not imply effectiveness. This is one of the first reviews to integrate peer-reviewed and grey literature on vibe coding under a single documented protocol. Future work should evaluate safeguard effectiveness, study session-level dynamics and long-term maintainability, and test vibe coding in production, data-intensive, and safety-critical settings.
Shahbaz Siddeeq, Muhammad Waseem, Kai-Kristian Kemell et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Preparing fine-tuning datasets for large language models (LLMs) commonly involves substantial manual effort, particularly in extracting, structuring, and validating data from unstructured sources. This study proposes a semi-automated, human-in-the-loop approach for generating fine-tuning question–answer (QA) pairs from PDF documents. The research investigates how unstructured textual content can be systematically transformed into validated QA data suitable for fine-tuning, while mitigating the risks associated with hallucinated or low-quality model outputs.The proposed system consists of a web-based architecture combining a React frontend with a Flask backend interfacing with the OpenAI API. Users provide a PDF document and a target page range, after which the system extracts text and generates candidate QA pairs. These candidates are presented for manual inspection, filtering, and refinement, prior to export in a structured JSON format compatible with fine-tuning pipelines.The results indicate that the proposed approach reduces the effort required for manual dataset construction while preserving data quality through mandatory human validation. The study highlights the effectiveness of hybrid automation workflows in accelerating fine-tuning dataset preparation without compromising reliability, and contributes design insights for human-centered tools supporting LLM customization.
Niklas Suvitie, Mika Saari, Pekka Abrahamsson· International Convention on...· 1 citation
Short-term Heart Rate Variability (HRV) forecasting could provide clinicians with actionable lead time for detecting autonomic dysfunction and adverse cardiac events. Consumer wearable devices generate fragmented, artifact-rich HRV signals that challenge conventional forecasting approaches. In this study, we evaluated the forecasting ability of three Time Series Foundation Models (TSFMs), TimesFM, Chronos, and MOIRAI, against traditional baselines (Mean, Exponential Smoothing, and Exponentially Weighted Moving Average) on real-world wearable data collected from 49 healthy individuals. To address data fragmentation, we introduce a variability-preserving imputation method that augments linear interpolation with locally adaptive stochastic noise, retaining physiological dynamics essential for accurate forecasting. The results show that TSFMs outperformed all baselines without fine-tuning, achieving average Mean Absolute Scaled Error (MASE) between 0.81 and 0.87 across TSFMs and both context lengths (32 and 64 time steps), with Chronos and TimesFM as the top models, though MOIRAI showed limited gains over baselines. With up to a 2-hour forecast horizon, the results establish a baseline for TSFMs'performance on a real-world dataset, highlighting domain-specific fine-tuning as a promising direction for clinical deployment.
Luukas Peräkylä, F. Sohrab, Ville Hautamäki et al.· 0 citations
Automated authoring of Gherkin Behavior-Driven Development (BDD) acceptance criteria remains a manual bottleneck in requirements engineering. This study investigates whether epic-organized LLM-generated Gherkin produces higher quality and coverage than requirement-aligned generation. We compare our Timeless (an epic-organized LLM pipeline) approach against a naive large language model (LLM) baseline on four requirements documents (107 requirements) from the PURE dataset. Evaluation covers structural metrics, automated requirement coverage via TF-IDF and dense embeddings, and blind expert assessment by four researchers. In our evaluation, the JSON-constrained pipeline produced structurally valid scenarios across all generated outputs, while the zero-shot baseline achieved 99% structural validity. Semantic coverage was comparable to the baseline, with Timeless achieving 94.3% semantic Requirement Coverage Rate compared with 92.9% for the baseline. TF-IDF produced lower coverage scores for the epic-organized output, suggesting that lexical metrics may miss coverage when scenarios paraphrase requirements at a higher level of abstraction. Expert raters prefer the epic-organized strategy on Correctness (4.61 vs 4.14), Executability (4.61 vs 4.07), and Completeness (4.31 vs 3.50). Overall, the results suggest that epic-organized generation can improve perceived Gherkin quality while maintaining comparable semantic coverage, although broader replication is needed before generalizing this finding.
Shahbaz Siddeeq, M. Abbasi, Jussi Rasku et al.· 0 citations
In agile software development, maintaining high-quality user stories is crucial, but also challenging. This study explores the use of large language models to automatically improve the user story quality in Austrian Post Group IT agile teams. We developed a reference model for an Autonomous LLM-based Agent System and implemented it at the company. The quality of user stories in the study and the effectiveness of these agents for user story quality improvement was assessed by 11 participants across six agile teams. Our findings demonstrate the potential of LLMs in improving user story quality, contributing to the research on AI role in agile development, and providing a practical example of the transformative impact of AI in an industry setting.
Zheying Zhang, M. Rayhan, Tomas Herda et al.· International Conference on...· 48 citations· ⚡4
Retrieval-Augmented Generation (RAG) systems are emerging as a key approach for grounding Large Language Models (LLMs) in external knowledge, addressing limitations in factual accuracy and contextual relevance. However, there is a lack of empirical studies that report on the development of RAG-based implementations grounded in real-world use cases, evaluated through general user involvement, and accompanied by systematic documentation of lessons learned. This paper presents five domain-specific RAG applications developed for real-world scenarios across governance, cybersecurity, agriculture, industrial research, and medical diagnostics. Each system incorporates multilingual OCR, semantic retrieval via vector embeddings, and domain-adapted LLMs, deployed through local servers or cloud APIs to meet distinct user needs. A web-based evaluation involving a total of 100 participants assessed the systems across six dimensions: (i) Ease of Use, (ii) Relevance, (iii) Transparency, (iv) Responsiveness, (v) Accuracy, and (vi) Likelihood of Recommendation. Based on user feedback and our development experience, we documented twelve key lessons learned, highlighting technical, operational, and ethical challenges affecting the reliability and usability of RAG systems in practice.
M. Hasan, Muhammad Waseem, Kai-Kristian Kemell et al.· EUROMICRO Conference on Soft...· 10 citations· ⚡1
Refactoring is a constant activity in software development and maintenance. Scale and maintain software systems are based on code refactoring. However, this process is still labor intensive, as it requires programmers to analyze the codebases in detail to avoid introducing new defects. In this research, we put forward a large language model (LLM)-based multi-agent system to automate the refactoring process on Haskell code. The objective of this research is to evaluate the effect of LLM-based agents in performing structured and semantically accurate refactoring on Haskell code. Our proposed multi-agent system based on specialized agents with distinct roles, including code analysis, refactoring execution, verification, and debugging. To test the effectiveness and practical applicability of the multi-agent system, we conducted evaluations using different open-source Haskell codebases. The results of the experiments carried out showed that the proposed LLM-based multi-agent system could average 11.03% decreased complexity in code, an improvement of 22.46% in overall code quality, and increase performance efficiency by an average of 13.27%. Furthermore, memory allocation was optimized by up to 14.57%. These results highlight the ability of LLM-based multi-agent in managing refactoring tasks targeted toward functional programming paradigms. Our findings hint that LLM-based multi-agent systems integration into the refactoring of functional programming languages can enhance maintainability and support automated development workflows.
Shahbaz Siddeeq, Muhammad Waseem, Z. Rasheed et al.· International Conference on...· 4 citations
Anomaly detection in smart power grids is a critical challenge due to the complexity, heterogeneity, and dynamic nature of sensor data streams. Existing one-class classification methods, particularly Subspace Support Vector Data Description (SVDD), have been extended to multimodal scenarios but often fail to fully exploit the structural dependencies across modalities, limiting their robustness in real-world applications. In this paper, we address this gap by proposing a generalized Multimodal Subspace Support Vector Data Description (MS-SVDD) model with graph-embedded regularization. The method projects data from multiple modalities into a shared low-dimensional subspace while preserving modality-specific structure through Laplacian regularizers. Our approach is evaluated on a three-modality dataset derived from smart grid event time series, using a dedicated preprocessing pipeline for constructing one-class classification training samples. The results demonstrate that our graph-embedded MS-SVDD improves robustness of event detection compared to conventional approaches, highlighting the potential of integrating graph priors with multimodal subspace learning for advancing anomaly detection in critical infrastructure. More broadly, this work contributes to the wider field of AI by illustrating how relational and structural information can be systematically embedded into one-class models, enabling robust learning under complex, high-dimensional, and multimodal conditions.
Thomas Debelle, F. Sohrab, Pekka Abrahamsson et al.· Scientific Reports· 1 citation
This paper presents MARARE, a real-time multi-agent system that transforms meeting dialogues into structured software requirements. One agent interacts with participants, while background agents extract and verify requirements collaboratively. Evaluation using the LLM-as-a-Judge method across five meetings (5–8 minutes each) shows a mean coverage of 80.0 ± 11.2 % (mean ± SD), semantic similarity of 0.86 ± 0.05, and hallucination rate of 14.3 ± 6.2 %. Preliminary results indicate performance differences across LLMs, suggesting that model choice influences coverage, consistency, and hallucination rates.
Malik Abdul Sami, Gessé Evangelista, Kai-Kristian Kemell et al.· AGENT@ICSE· 0 citations
What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration. The post GigaPath-Flash and GigaTIME-Flash: Toward population-scale discovery with efficient pathology foundation models appeared first on Microsoft Research.