SUITE5 and the coordination view on the expected scientific and technical takeaways
AI DAPT aims to assist stakeholders from various industries in overcoming data-related challenges that may come across their AI transformation. Introducing AI-assisted automation with the human-in-the-loop (HITL) in data AI pipelines, AI DAPT will facilitate the design, execution, observability and management of data-AI operations within a novel AI Ops system that is continuously learning and adapting based on context.
Transforming Data AI Pipelines
This cross-cutting AIOps intelligent lifecycle framework that bridges business, legal, ethics and scientific aspects will utilise novel techniques, and will push the boundaries to go beyond state-of-the-art in research paths that fall under two main areas:
Area #1 – Data pipelines for robust and trustworthy AI, including data management & synthetic data generation
Introducing automation in data preprocessing for AI: AI DAPT introduces automated, intelligent ‘Data for AI’ pipelines that will incorporate the human-AI teaming approach along all steps of the data lifecycle within an AI system. AI DAPT will achieve the balance between automation and human control, and speed-up data operations that until now needed immense human effort.
Introducing Explainability: Injection of sophisticated observability techniques and Explainable AI methods in all steps of the data pipelines will allow the end-users understand their data, comprehend the results of their Data pipelines and ultimately create optimized datasets for their AI operations.
Introducing Data Valuation and Synthetic Data Generation: Through decision-based data valuation algorithms for the assessment data, under the prism of fitness-for-use in the designated AI applications and for the detection of underlying bias; AI DAPT will assist users in improving their datasets and provide synthetic data generation methods to augment or replace real data as a method for improved AI models, sensitive data protection, and mitigation of bias and imbalance.
Area #2 – AI pipelines including Hybrid science-ML, Efficient AI-Ops, and Adaptive AI
Introducing Hybrid Science-Guided AI: Bringing together the best of two disparate worlds until now, AI and. science-based knowledge and modelling, the AI-DAPT pipelines will achieve more accurate and scientifically consistent predictions. First-principles models from the health, energy, manufacturing and robotics domains will be investigated and employed.
Introducing Continuous model Improvement with the human-in-the-loop: Model observability techniques and XAI will facilitate users continuously monitor and evaluate their models’ performance, while adaptive AI techniques intertwined within the AI pipeline orchestration and execution flow will allow informed post deployment model retraining and optimisation without disrupting the core operation of the pipeline and requiring the minimum re-configuration effort from the side of the data scientist.
Scientific and Technical Outcomes
A multitude of key technical and scientific results will incorporate the advancements made within the AI DAPT project under the above discussed areas and will be the AI DAPT’s contribution to the research community.
- In depth state of the art analysis on adjacent research areas: From research and technology, up to market and legal/ethics perspectives, comprehensive landscape analysis will be performed and published through the relevant AI DAPT public deliverables and other publications constituting a consolidation of current state-of-the art for other researchers to utilise and extend in their own endeavors.
- Fusion, experimentation and advancement of existing techniques: Existing methods in the areas of data curation, hybrid science AI, XAI, human-in-the-loop data and AI operations, synthetic data generation, data and model observability and more, will be tested for their applicability within AI DAPT. Advancements in existing techniques are expected, to address known limitations and cover the requirements of the AI DAPT demonstrators.
- Automated and Interoperable Data AI Pipelines: Pipeline blueprints for end-to-end configuration of data and AI operations will be designed, targeted both for the AI DAPT Integrated framework but also for usage within external AI solutions.
- Validated AI DAPT individual Services and Integrated AI Ops Framework: The AI DAPT components, that will be integrated under the AI DAPT Framework and orchestrated via the appropriate horizontal services, will bring into life the AI DAPT Data AI Pipelines. The technological developments of AI DAPT will be validated and verified in four pilots from four strategic industries (energy, health, manufacturing, robotics) in order to assure the high quality and relevance of the developed solutions for the end-user.
- AI-DAPT Technology Radar, scientific Publications and events: AI DAPT will adopt an open science approach in order to maximize the visibility of its results and advancements. Under this prism, the AI DAPT technology radar will constantly monitor and map the research and technological landscape in the domains that are pertinent to the data and AI end-to-end operations. Additionally, new findings and advancements under the frame of the research and technological work of AI DAPT will be published through high reputation, open access conferences and journals to achieve wide diffusion of knowledge to the research community.
The SUITE5 Team
SUITE5 is a data intelligence solutions SME, based in Limassol, Cyprus. SUITE5 brings in AI DAPT its expertise in data management technologies, analytics and AI, and trusted data sharing operations across multi-faceted sectors and applications (Energy, Aviation, Manufacturing, etc.), as well as vast experience from the technical coordination in various projects (HEU – SYNERGIES Energy Data Space project, H2020-XMANAI project, H2020-SYNERGY project, H2020-ICARUS project).
SUITE5, as the Technical Coordinator of AI DAPT, will orchestrate and supervise the research and development activities of the project in order to ensure the technical excellence and scientific robustness of the AI DAPT solutions that will facilitate transformation across diverse domains through data-centric AI pipelines.
Furthermore, SUITE5 will onboard in the AI DAPT project its strong R&D department conducting basic and applied research in AI and XAI, for the implementation of various building blocks of the AI DAPT Data/AI Lifecycle, including data generation for AI, model observability, XAI techniques and the configuration and execution of data pipelines.
Our key personnel
Nefeli Bountouni is a Senior Project Manager at Suite5, specializing in the management of EU-funded research and innovation projects with a strong focus on digital transformation, sustainability, and data-driven solutions. She holds a Bachelor’s degree in International and European Economic Studies from Athens University of Economics and Business, and a Master’s degree in Environmental Governance and Sustainable Development from Panteion University of Social and Political Sciences. With extensive experience coordinating large-scale R&I initiatives, Nefeli is particularly interested in digital innovation, climate action, and policy development. Her work is centered around delivering impactful results in EU-funded projects across multiple sectors, including energy, environment, and digital technologies.
Sotiris Koussouris is the Director of Innovation at Suite5, where he leads the strategic direction of the company’s R&D initiatives. He holds a Diploma in Electrical and Computer Engineering from the National Technical University of Athens (NTUA) and a PhD in Computer Science from the same institution. With over 15 years of experience, Sotiris has an extensive background in leading EU-funded projects in the areas of data management, AI, and digital transformation. His research interests include large-scale data analytics, cloud computing, and digital platforms for the industrial and public sectors. Sotiris has coordinated multiple high-impact projects and published several papers in international journals and conferences.