ABOUT ME

Originally from India, I am currently in the final stage of my PhD in Artificial Intelligence and Machine Learning at the University of Limerick, Ireland, funded by the Science Foundation Ireland Centre for Research Training in AI (SFI CRT-AI). My work focuses on building modular, reusable, and reproducible machine learning pipelines for applied use cases in healthcare, edge computing, and data analytics, with emphasis on stakeholder usability, interpretability, and performance. Prior to this, I completed an M.Sc. in Data Analytics from the National College of Ireland with First Class Honours, where I specialized in machine learning, statistical modelling, and scalable data workflows.

I also hold an M.Sc. in Physics with a specialization in Theoretical Astrophysics & Cosmology from the University of Zurich, and a B.Sc. (Honours) in Physics from the University of Delhi. These academic foundations have given me the analytical and mathematical depth to approach problems systematically. I am proficient in Python and familiar with SQL and Git; I have worked with frameworks such as scikit-learn, TensorFlow, Keras, and PyTorch for model development, and explored tools like Docker, MLFlow, and GitHub Actions in the context of reproducibility and deployment. I’ve also contributed to the design and evaluation of low-code/no-code AI platforms (CINCO de Bio, Pyrus, DIME), aimed at enabling non-technical users to build and reuse ML pipelines.

Professionally, I have served as an AI/ML Engineer (PhD) and Applied Researcher (PhD) at the University of Limerick, where I worked on full-cycle ML pipelines, real-time edge analytics, and experimental MLOps foundations. I have also delivered lectures and provided technical mentoring in my role as Academic Teaching & Mentoring faculty, covering subjects like AI/ML, data analytics, and software testing. In parallel with my research, I’ve authored 10+ peer-reviewed publications in conferences and journals, with contributions in model interpretability, workflow automation, and domain-specific AI. I’m also passionate about open-source learning and enjoy building clean, testable code and lightweight web projects to improve my HTML/CSS skills.

---

Work Experience

AI/ML Engineer – Pipeline Development & AI Experimentation (PhD)

University of Limerick (Sep 2021 – Present)
- Designed and implemented modular ML pipelines using Python and SQL for structured and semi-structured datasets.
- Focused on reproducibility, traceability, and Git-based version control.
- Evaluated classification and summarisation workflows for research reporting and stakeholder-aligned outcomes.
- Collaborated with engineers and domain experts to produce dashboard summaries and technical insights.

Applied Researcher – Edge AI & MLOps Foundations (PhD)

University of Limerick (Jan 2023 – Dec 2023)
- Built real-time data pipelines on embedded devices (Raspberry Pi, Thingy:53) for latency-sensitive health monitoring.
- Contributed to early-stage MLOps workflows using configuration-driven metadata tracking and automated evaluation pipelines.
- Evaluated tools like MLFlow, Docker, and GitHub Actions to assess reproducibility, deployment readiness, and integration feasibility.

Academic Teaching & Mentoring Experience

University of Limerick & National College of Ireland (Jan 2021 – May 2023)
- Delivered lectures, lab sessions, and technical mentoring in AI/ML, Software Testing, Data Analytics, and Programming.
- Supervised student-led projects involving classification pipelines, data visualisation, Python-based dashboards, and validation workflows.
- Reinforced principles of reproducibility, modularity, version control, and Agile-style development in collaborative coursework.
- Modules taught or supported include:
- Theory and Practice of Advanced AI Ecosystems (UL)
- Software Testing and Inspection, Problem Solving with Computers (UL)
- Artificial Intelligence, Data Mining & ML, Data Visualisation, SQL & Databases, Algorithms & Advanced Programming (NCI)

---

Education

PhD in Artificial Intelligence and Machine Learning (AI/ML) (Sep 2021 – May 2025)

University of Limerick, Limerick, Ireland

• Fully-funded by Science Foundation Ireland Centre for Research and Training in Artificial Intelligence (SFI CRT-AI).
• Modules: Machine Learning, Optimization and Constraint Programming, Visual Media Processing, Reinforcement Learning and Personalisation, Natural Language Processing, Research Integrity, Digital Research Management, Research Networking: Developing an Academic Profile, Planning Research & Publication, Research Ethics, Developing Ideas & Arguments: Writing into Academic Communities.

M.Sc. in Data Analytics (DA) (Jan 2020 – Feb 2021)

National College of Ireland, Dublin, Ireland

• Grades: 77.1%, 1.1 grade, First Class Honours.
• Modules: Data Mining and Machine Learning-I (DMML-1), Database and Analytics Programming, Business Intelligence and Business Analytics, Statistics for Data Analytics, Data Mining and Machine Learning-II (DMML-2), Modelling Simulation and Optimization, Domain Application of Predictive Analytics, Research in Computing, and Data Governance and Ethics.

M.Sc. in Physics - Theoretical Astrophysics & Cosmology (Sep 2017 – Oct 2019)

University of Zurich, Zurich, Switzerland

• Grades: 4.5/6, 2.1 grade.
• Modules: Theoretical Astrophysics, Theoretical Cosmology, Astro-Particle Physics-I & II, Introduction to Astrobiology, Extra-Solar Planets, The Sun & the Planets, Planet Formation.

B.Sc. (Honours) in Physics (Jul 2014 – Jul 2017)

University of Delhi, New Delhi, India

• Grades: 77.64%, 1.1 grade, First Class Honours.
• Relevant Modules: Mathematics (Analysis & Statistics), Mathematical Physics (Linear Algebra, Calculus, Linear Programming), Numerical Analysis (C++), Microprocessor Programming, Digital Electronics.

---

Publications

Low-code internet of things application development for edge analytics

• Publication: IFIP International Internet of Things Conference, Springer International Publishing (Conference)
• Date Published: 19.10.2022
• URL: Read More
• Citation: H.A.A. Chaudhary, I. Guevara, J. John, A. Singh, T. Margaria, and D. Pesch, Low-Code Internet of things application development for edge Analytics, IFIP advances in information and communication technology. pp. 293–312 (2022).
• Abstract: Internet of Things (IoT) applications combined with edge analytics are increasingly developed and deployed across a wide range of industries by engineers who are non-expert software developers. In order to enable them to build such IoT applications, we apply low-code technologies in this case study based on Model Driven Development. We use two different frameworks: DIME for the application design and implementation of IoT and edge aspects as well as analytics in R, and Pyrus for data analytics in Python, demonstrating how such engineers can build innovative IoT applications without having the full coding expertise. With this approach, we develop an application that connects a range of heterogeneous technologies: sensors through the EdgeX middleware platform with data analytics and web based configuration applications. The connection to data analytics pipelines can provide various kinds of information to the application users. Our innovative development approach has the potential to simplify the development and deployment of such applications in industry.

Model-driven engineering in digital thread platforms: a practical use case and future challenges

• Publication: International Symposium on Leveraging Applications of Formal Methods, Springer Nature Switzerland (Conference)
• Date Published: 17.10.2022
• URL: Read More
• Citation: H.A.A. Chaudhary, I. Guevara, J. John, A. Singh, A. Ghosal, D. Pesch, and T. Margaria, Model-Driven Engineering in Digital thread Platforms: a practical use case and future challenges, Lecture notes in computer science. pp. 195–207 (2022).
• Abstract: The increasing complexity delivered by the heterogeneity of the cyber-physical systems is being addressed and decoded by edge technologies, IoT development, robotics, digital twin engineering, and AI. Nevertheless, tackling the orchestration of these complex ecosystems has become a challenging problem. Specially the inherent entanglement of the different emerging technologies makes it hard to maintain and scale such ecosystems. In this context, the usage of model-driven engineering as a more abstract form of glue-code, replacing the boilerplate fashion, has improved the software development lifecycle, democratising the access to and use of the aforementioned technologies. In this paper, we present a practical use case in the context of Smart Manufacturing, where we use several platforms as providers of a high-level abstraction layer, as well as security measures, allowing a more efficient system construction and interoperability.

Efficient Model-Driven Prototyping for Edge Analytics

• Publication: MDPI Electronics 2023 (Journal)
• Date Published: 14.09.2023
• URL: Read More
• Citation: H.A.A. Chaudhary, I. Guevara, A. Singh, A. Schieweck, J. John, T. Margaria, and D. Pesch, Efficient Model-Driven prototyping for edge analytics, Electronics. vol. 12, no. 18, pp. 3881, 2023.
• Abstract: Software development cycles in the context of Internet of Things (IoT) applications require the orchestration of different technological layers, and involve complex technical challenges. The engineering team needs to become experts in these technologies and time delays are inherent due to the cross-integration process because they face steep learning curves in several technologies, which leads to cost issues, and often to a resulting product that is prone to bugs. We propose a more straightforward approach to the construction of high-quality IoT applications by adopting model-driven technologies (DIME and Pyrus), that may be used jointly or in isolation. The presented use case connects various technologies: the application interacts through the EdgeX middleware platform with several sensors and data analytics pipelines. This web-based control application collects, processes and displays key information about the state of the edge data capture and computing that enables quick strategic decision-making. In the presented case study of a Stable Storage Facility (SSF), we use DIME to design the application for IoT connectivity and the edge aspects, MongoDB for storage and Pyrus to implement no-code data analytics in Python. We have integrated nine independent technologies in two distinct Low-code development environments with the production of seven processes and pipelines, and the definition of 25 SIBs in nine distinct DSLs. The presented case study is benchmarked with the platform to showcase the role of code generation and the reusability of components across applications. We demonstrate that the approach embraces a high level of reusability and facilitates domain engineers to create IoT applications in a low-code fashion.

CNN-based Human Activity Recognition on Edge Computing Devices

• Publication: IEEE COINS 2023 (Conference)
• Date Published: 23.07.2023
• URL: Read More
• Citation: A. Singh, T. Margaria, and F. Demrozi, CNN-based Human Activity Recognition on Edge Computing Devices, 2023 IEEE International Conference on Omni-layer Intelligent Systems (COINS), Jul. 2023, pp. 1–4.
• Abstract: Human Activity Recognition (HAR) is a research area that involves wearable devices integrating inertial and/or physiological sensors to classify human actions and status across various application domains, such as healthcare, sports, industry, and entertainment. However, executing HAR algorithms on remote devices or the cloud can lead to issues such as latency, bandwidth requirements, and energy consumption. Transitioning towards Edge HAR can be a more effective and versatile solution, overcoming the challenges of traditional HAR techniques. We present a novel HAR model for computation on edge devices: we design a Convolutional Neural Network (CNN) Deep Learning approach and compare its performance with cloud-computing HAR models. The paper is accompanied by a self-collected dataset. The experiments on this dataset demonstrate that the proposed edge computing model achieves promising results ( ≥ 92 %) in terms of Precision, Recall, and Fl-score. Furthermore, the model exhibits significantly reduced latency, with only 117 ms, and utilizes minimal memory, with a peak of 18.8 Kb RAM and 956 Kb Flash memory.

IDPP: Imbalanced Datasets Pipelines in Pyrus

• Publication: ECBS 2023 (Conference)
• Date Published: 16.10.2023
• URL: Read More
• Citation: A. Singh and O. Minguett, IDPP: Imbalanced Datasets Pipelines in Pyrus, Engineering of Computer-Based Systems, J. Kofroň, T. Margaria, and C. Seceleanu, Eds., in Lecture Notes in Computer Science. Cham: Springer Nature Switzerland, 2024, pp. 60–69.
• Abstract: We showcase and demonstrate IDPP, a Pyrus-based tool that offers a collection of pipelines for the analysis of imbalanced datasets. Like Pyrus, IDPP is a web-based, low-code/no-code graphical modelling environment for ML and data analytics applications. On a case study from the medical domain, we solve the challenge of re-using AI/ML models that do not address data with imbalanced class by implementing ML algorithms in Python that do the re-balancing. We then use these algorithms and the original ML models in the IDPP pipelines. With IDPP, our low-code development approach to balance datasets for AI/ML applications can be used by non-coders. It simplifies the data-preprocessing stage of any AI/ML project pipeline, which can potentially improve the performance of the models. The tool demo will showcase the low-code implementation and no-code reuse and repurposing of AI-based systems through end-to end Pyrus pipelines.

Binary Decision Diagrams and Composite Classifiers for Analysis of Imbalanced Medical Datasets

• Publication: ECEASST 2023 (Journal)
• Date Published: 06.10.2023
• URL: Read More
• Citation: A. Singh, O. Minguett, and T. Margaria, Binary Decision Diagrams and Composite Classifiers for Analysis of Imbalanced Medical Datasets, Electronic Communications of the EASST, vol. 82, Oct. 2023.
• Abstract: Imbalanced datasets pose significant challenges in the development of accurate and robust classification models. In this research, we propose an approach that uses Binary Decision Diagrams (BDDs) to conduct pre-checks and suggest appropriate resampling techniques for imbalanced medical datasets as the application domain where we apply this technology is medical data collections. BDDs provide an efficient representation of the decision boundaries, enabling interpretability and providing valuable insights. In our experiments, we evaluate the proposed approach on various real-world imbalanced medical datasets, including Cerebralstroke dataset, Diabetes dataset and Sepsis dataset. Overall, our research contributes to the field of imbalanced medical dataset analysis by presenting a novel approach that uses BDDs and composite classifiers in a low-code/no-code environment. The results highlight the potential for our method to assist healthcare professionals in making informed decisions and improving patient outcomes in imbalanced medical datasets.

Edge IoT prototyping using Model-Driven representations: A use-case for Smart Agriculture

• Publication: MDPI Sensors 2023 (Journal)
• Date Published: 12.01.2024
• URL: Read More
• Citation: I. Guevara, S. Ryan, A. Singh, C. Brandon, and T. Margaria, Edge IoT Prototyping Using Model-Driven Representations: A Use Case for Smart Agriculture, Sensors, vol. 24, no. 2, Art. no. 2, Jan. 2024.
• Abstract: Industry 4.0 is positioned at the junction of different disciplines, aiming to re-engineer processes and improve effectiveness and efficiency. It is taking over many industries whose traditional practices are being disrupted by advances in technology and inter-connectivity. In this context, enhanced agriculture systems incorporate new components that are capable of generating better decision making (humidity/temperature/soil sensors, drones for plague detection, smart irrigation, etc.) and also include novel processes for crop control (reproducible environmental conditions, proven strategies for water stress, etc.). At the same time, advances in model-driven development (MDD) simplify software development by introducing domain-specific abstractions of the code that makes application development feasible for domain experts who cannot code. XMDD (eXtreme MDD) makes this way to assemble software even more user-friendly and enables application domain experts who are not programmers to create complex solutions in a more straightforward way. Key to this approach is the introduction of high-level representations of domain-specific functionalities (called SIBs, service-independent building blocks) that encapsulate the programming code and their organisation in reusable libraries, and they are made available in the application development environment. This way, new domain-specific abstractions of the code become easily comprehensible and composable by domain experts. In this paper, we apply these concepts to a smart agriculture solution, producing a proof of concept for the new methodology in this application domain to be used as a portable demonstrator for MDD in IoT and agriculture in the Confirm Research Centre for Smart Manufacturing. Together with model-driven development tools, we leverage here the capabilities of the Nordic Thingy:53 as a multi-protocol IoT prototyping platform. It is an advanced sensing device that handles the data collection and distribution for decision making in the context of the agricultural system and supports edge computing. We demonstrate the importance of high-level abstraction when adopting a complex software development cycle within a multilayered heterogeneous IT ecosystem.

CINCO de Bio: A Future Internet Platform for Domain-Specific Workflows that Leverage AI for Biomedical Research

• Publication: MDPI BioMedInformatics (Journal)
• Date Published: 09/2024
• URL: Read More
• Citation: C. Brandon, S. Boßelmann, A. Singh, S. Ryan, A. Schieweck, E. Fennell, B. Steffen, and T. Margaria, CINCO de Bio: A Low-Code Platform for Domain-Specific Workflows for Biomedical Imaging Research, BioMedInformatics. vol. 4, no. 3, pp. 1865–1883, 2024.
• Abstract: Background: In biomedical imaging research, experimental biologists generate vast amounts of data that require advanced computational analysis. Breakthroughs in experimental techniques, such as multiplex immunofluorescence tissue imaging, enable detailed proteomic analysis, but most biomedical researchers lack the programming and Artificial Intelligence (AI) expertise to leverage these innovations effectively. Methods: Cinco de Bio (CdB) is a web-based, collaborative low-code/no-code modelling and execution platform designed to address this challenge. It is designed along Model-Driven Development (MDD) and Service-Orientated Architecture (SOA) to enable modularity and scalability, and it is underpinned by formal methods to ensure correctness. The pre-processing of immunofluorescence images illustrates the ease of use and ease of modelling with CdB in comparison with the current, mostly manual, approaches. Results: CdB simplifies the deployment of data processing services that may use heterogeneous technologies. User-designed models support both a collaborative and user-centred design for biologists. Domain-Specific Languages for the Application domain (A-DSLs) are supported through data and process ontologies/taxonomies. They allow biologists to effectively model workflows in the terminology of their field. Conclusions: Comparative analysis of similar platforms in the literature illustrates the superiority of CdB along a number of comparison dimensions. We are expanding the platform’s capabilities and applying it to other domains of biomedical research.

Enhancing Decision-Making for imbalanced medical datasets using BDDs and Low-Code/No-Code

• Publication: IT Professional (Journal/Magazine)
• Date Published: 09/2024
• URL: Read More
• Citation: A. Singh and T. Margaria, Enhancing Decision-Making for Imbalanced Medical Datasets Using BDDs and Low-Code/No-Code, in IT Professional, vol. 26, no. 5, pp. 92-98, 09-10 2024.
• Abstract: Imbalanced datasets pose a challenge wherever accurate predictions are essential. This paper explores using low-code/no-code platforms, such as Pyrus and ADD-Lib, to apply data resampling techniques and binary decision diagrams for more accessible and reliable ML workflows. Tested on three medical datasets, these techniques improve model performance by addressing class imbalances. The integration of resampling and formal methods enhances prediction accuracy while making ML tools more accessible to professionals, enabling better decision-making in critical applications. The techniques are applicable to any domain, not just in healthcare.

Model driven development for AI-Based healthcare Systems: a review

• Publication: LNCS, Springer Nature Switzerland (Conference)
• Date Published: 10/2024
• URL: Read More
• Citation: C. Brandon, A. Singh, and T. Margaria, Model driven development for AI-Based healthcare Systems: a review, Lecture Notes in Computer Science. pp. 245–265, 2024.
• Abstract: We review our experience with integrating Artificial Intelligence (AI) into healthcare systems following the Model-Driven Development (MDD) approach. At a time when AI has the potential to instigate a paradigm shift in the health sector, better integrating healthcare experts in the development of these technologies is of paramount importance. We see MDD as a useful way to better embed non-technical stakeholders in the development process. The main goal of this review is to reflect on our experiences to date with MDD and AI in the context of developing healthcare systems. Four case studies that fall within that scope but have different profiles are introduced and summarised: the MyMM application for Multiple Myeloma diagnosis; CNN-HAR, that studies the ability to do AI on the edge for IoT-supported human activity recognition; the HIPPP web based portal for patient information in public health; and Cinco de Bio, a new model driven platform used for the first time to support a better cell-level understanding of diseases. Based on the aforementioned case studies we discuss the characteristics, the challenges faced and the postive outcomes achieved.