Summer School
Cybersecurity, Threat Intelligence,
IoT, Blockchain & Computational Modeling

Morning Keynote, Daily opening lecture

Hack the Mindset: Innovation, Security, and Tomorrow’s World

Inside the mind of a cyber attacker: The psychology behind digital threats

Cyber attacks are not driven by technology alone—they are driven by people, motives, and psychology. To effectively defend digital systems, we must first understand the human mindset behind malicious behavior. This presentation explores the psychology of cyber attackers, examining the motivations, decision-making processes, personality traits, and social engineering tactics that shape modern digital threats. From financial gain and ideological agendas to curiosity, revenge, and the pursuit of status, attackers operate with diverse intentions and methods. By analyzing how they think, plan, manipulate, and exploit trust, participants will gain valuable insight into the human factors of cybersecurity. Understanding the attacker’s mindset is a critical step toward building stronger defenses, improving awareness, and creating a more resilient digital future.

Cyber Geopolitics: HUMINT, SIGINT, GEOINT in Military and Diplomatic Operations

This talk explores the growing intersection between cybersecurity, geopolitics, and intelligence disciplines in an era defined by digital competition and hybrid threats. Titled Cyber Geopolitics: HUMINT, SIGINT, GEOINT in Military and Diplomatic Operations, the session examines how Human Intelligence (HUMINT), Signals Intelligence (SIGINT), and Geospatial Intelligence (GEOINT) are increasingly integrated to support national security decision-making, military planning, and diplomatic strategy. As cyber operations become central to statecraft, intelligence agencies and defense institutions rely on multi-source intelligence fusion to identify emerging threats, attribute hostile actions, protect critical infrastructure, and shape strategic responses. The talk highlights how cyber espionage, information warfare, satellite surveillance, communications interception, and human-source networks now operate within a common strategic framework. Special attention is given to the relevance of these developments for emerging technologies such as the Internet of Things (IoT) and Blockchain ecosystems. Expanding IoT infrastructures create new intelligence opportunities and attack surfaces, while Blockchain technologies introduce both resilience mechanisms and novel security challenges. Understanding how intelligence capabilities interact with these technologies is essential for future cybersecurity professionals. Designed for participants of the Summer School on Cybersecurity, IoT and Blockchain, the talk combines real-world case studies, strategic analysis, and forward-looking insights to help students understand how cyber power, intelligence operations, and global security dynamics are reshaping the modern threat landscape.

Reading Plasma Light: Futuristic spectroscopy for light analyses

This presentation focuses on light-based plasma diagnostics as a gateway to controlling the chemistry of cold atmospheric-pressure plasma jets. Using optical emission spectroscopy, the study analyzes how molecular nitrogen and oxygen injection modifies the light emitted by key plasma species. By comparing conventional premixed injection with a Y-shaped in-discharge injection geometry, the work shows how spectral signatures reveal changes in plasma stability, radical formation, electron density, and rotational temperature. The Y-DBD configuration enables stable operation at much higher molecular gas fractions, while maintaining strong and balanced emission from reactive species. In contrast, the I-DBD configuration offers selective enhancement of NO emission but is limited by rapid plasma destabilization.

National Cybersecurity Ecosystem

Romania’s national cybersecurity ecosystem is built on cooperation between government institutions, critical infrastructure operators, private industry, academia, and international partners, with the National Directorate of Cybersecurity (DNSC) serving as a central pillar. As the national authority for civilian cyberspace, DNSC coordinates prevention, incident response, awareness, resilience programs, and strategic policy development to strengthen the country’s digital security posture. Through public-private collaboration, education initiatives, threat intelligence sharing, and alignment with European cybersecurity frameworks, Romania continues to mature its capabilities against evolving cyber threats. This presentation highlights how a coordinated national ecosystem can enhance trust, innovation, and resilience in an increasingly interconnected world.

Artificial Intelligence (AI) in Veterinary Medicine: From Assistance to Limitations

Artificial intelligence is transforming veterinary medicine through applications already operational in medical imaging, assisted diagnostics, clinical workflow management, and owner-facing digital platforms. AI algorithms have demonstrated superior screening performance — including hyperadrenocorticism detection at 96.3% sensitivity and 97.2% specificity — while smart stethoscopes now enable early detection of hypertrophic cardiomyopathy in cats, and mobile applications bring AI-assisted triage directly to pet owners. Yet technology alone does not define this transformation: it is shaped by data quality, institutional trust, and the ethical frameworks governing its use. This presentation examines the motivations behind AI adoption in veterinary practice, the decision-making logic of algorithmic systems, and the risks that emerge when clinical judgment is delegated to opaque models — including hallucinations, algorithmic bias, and overdiagnosis. A case study on AI integration in intensive carp aquaculture illustrates both the precision gains and the ethical tensions inherent in automated veterinary decision-making. Understanding how AI systems think, fail, and influence clinical outcomes is a critical step toward building responsible frameworks, improving practitioner awareness, and ensuring that technological progress genuinely serves animal welfare.

Lunch break, Coffee Time!

Monitoring and digital assistance system for type 1 diabetes management

Maintaining optimal blood glucose levels requires accurate insulin dosage, proper carbohydrate counting, regular physical activity, and effective stress management. This paper presents a software application designed to support individuals with type 1 diabetes in managing these aspects effectively. The application provides functionality for calculating carbohydrate content of meals based on user-input data, helping users make more informed nutritional decisions. In addition, it improves user safety by enabling real-time monitoring of blood glucose levels and generating alerts when critical thresholds are reached. These alerts ensure that relevant information about the user’s condition can be quickly accessed in emergency situations. Furthermore, the application facilitates communication between individuals with type 1 diabetes through a social component that allows connection with trusted users, promoting mutual support and timely assistance. By integrating monitoring, communication, and alerting mechanisms, the system aims to reduce user stress and increase confidence in daily disease management. Overall, the proposed solution enhances self-management, awareness, and responsiveness in situations where rapid intervention may be necessary.

AI Platform for Real-Time Detection of Cyber Threats

The project addresses system vulnerabilities to modern cyberattacks, that start to use Artificial Intelligence to bypass traditional security layers. Since the data volume is exceeding human capacity to monitor network traffic logs, the proposed platform uses Artificial Intelligence to detect threats such as DDoS , Flood , Port Scan. Unlike slow manual log analysis, the platform has an Autoencoder model trained based on a normal activity dataset and reacts in real time. The system also features a modern graphical dashboard to help specialists in cybersecurity understand the potential risks. The goal is to effectively protect the data of online environments that get more dangerous by day.

Card Component Gallery

Card Component Gallery is a modern full-stack web application designed to showcase reusable UI card components in a clean and interactive interface. The project demonstrates how different types of cards (such as profile cards, product cards, and pricing plans) can be dynamically generated and displayed within a responsive layout. The front-end of the application is built using HTML, CSS, and JavaScript, focusing on a minimalist design and a smooth user experience. The interface allows users to visualize multiple card types in a structured gallery format, with filtering options that enhance usability and organization. On the back-end, the application uses Node.js and Express to create a lightweight server that provides data through a custom API endpoint. The card information is stored in a JSON file, making the content easy to manage and extend. Through asynchronous requests (fetch), the front-end retrieves this data and renders the components dynamically, transforming the application from a static page into a data-driven experience. This project highlights essential web development concepts such as separation of concerns, client-server communication, dynamic rendering, and reusable component design. It also serves as a foundation for more advanced applications that could integrate databases, authentication systems, or real-time features.

Modern Styled Login/Register Page with Smooth UX/UI and Security Features

This project involves the creation of a modern, visually appealing, and highly interactive Login and Register page. The main focus was placed on delivering an excellent User Experience (UX) and User Interface (UI) through elegant design, smooth transitions, animations, and intuitive navigation between login and registration forms. The interface features a dynamic modal popup system, real-time visual field validation, and polished animations that enhance user engagement. A looping background video was integrated to create an immersive and contemporary aesthetic. Special attention was given to responsive design, color harmony, typography, and micro-interactions. On the technical side, the frontend was developed using HTML5, CSS3, and vanilla JavaScript. For security, Google reCAPTCHA v2 was implemented to protect the registration form against spam and automated bots. The system is supported by a backend built with Node.js, Express.js, and a MySQL database (managed via MySQL Workbench), ensuring secure user data storage with password hashing using bcrypt. This project successfully combines modern web design principles with functional backend development, resulting in a beautiful, responsive, and secure authentication interface.

Authentication System / Stop Building Auth from Scratch — This Complete System Does It All

This project presents a complete authentication system that allows users to register, log in, and log out securely. It includes essential features such as password reset and email verification. Security is ensured through the use of JWT tokens, the bcrypt algorithm for password hashing, as well as rate limiting and security headers. The goal of the project is to provide a modern, secure, and easy‑to‑integrate solution for managing authentication in a web application.

AI-Based Predictive Maintenance and Intelligent Alarm Management in Maritime Navigation Systems

This work explores the integration of Artificial Intelligence (AI) into the predictive maintenance and management of maritime navigation systems. Modern vessels rely on complex electronic equipment such as ECDIS, RADAR, GPS, and AIS, whose reliability is critical for operational safety and efficiency. Traditional maintenance approaches, based on scheduled or reactive strategies, are insufficient in dynamic maritime environments. The proposed approach leverages Machine Learning (ML) techniques for anomaly detection, signal processing, and predictive analysis using data collected from onboard navigation equipment. The system enables early fault detection, intelligent alarm classification, and adaptive maintenance strategies. Simulation models implemented in Python demonstrate the feasibility of time-series prediction and pattern recognition using neural networks. The results indicate significant improvements in system reliability, reduction of downtime, and enhanced decision support for maritime operators. This work contributes to the development of smart shipping systems and supports the transition toward autonomous and digitally enhanced maritime operations.

Sending-or-Not-Sending Quantum Key Distribution with 2 Phase Matching

The discussion will present the performance of a hybrid quantum distribution protocol 12 that integrates the Sending-or-Not-Sending and Phase-Matching techniques of Twin-Field QKD protocols, thus exploiting all detection events, both single-sided (send/not-send) and double-sided (send-send) events. The goal was to fully utilize quantum resources, while increasing the rate and distance. 16 The comparative study was done in relation to the standard SNS-TF-QKDP and PM-17 QKD protocols (with discrete phase randomization, M=16 phase slices), the numerical simulation results 18 showing an improvement in the secret rate for the proposed model: a relative gain of over 19 50% over SNS over the distance range of 100-300 km and over 36% at 400 km, having a major competitive advantage over PM (a 3 times higher secret key rate). The analysis 21 also highlighted that an optimized SNS protocol becomes superior to all these protocols 22 starting from a distance of 333 km, thus becoming the best choice for long distances, 23 while the proposed hybrid protocol SNS-PM (optimized on the SNS component) remains the preferred variant for short and medium distances.

Secure by Design: A Practical, Scalable, and Cost-Free Serverless Web Architecture

This work presents the design and implementation of a practical, scalable, and cost-free serverless web architecture that can be readily deployed in real-world scenarios. The proposed system delivers a fully functional application, including a multi-page web interface, a contact system, and an administrative panel. The architecture leverages modern serverless technologies to minimize operational costs while maintaining performance and scalability. Security is treated as a core design principle rather than an afterthought. The system incorporates essential protection mechanisms such as input validation, secure authentication, and controlled database access. In addition, a dedicated demonstration module highlights the contrast between insecure and secure backend practices, emphasizing common vulnerabilities and their mitigation. The results demonstrate that a production-ready, secure web platform can be built with near-zero infrastructure costs, making this approach a viable and accessible solution for real-world deployment.

Automated code security and remediation

Traditional security reviews are often reactive and take hours. This talk demonstrates a proactive approach using AI-driven agents to catch vulnerabilities during the development phase. Learn how to use OpenCode and Git automation to create a self-contained review system that analyzes commits and provides automated feedback to developers.

Vibe hacking - we are not ready

As the AI race is reaching new levels, the cyberscurity industry will be more and more invloved with addressing the new threat models that occur when LLMs get used to test the defence of the increasingly complex digital infrastructures. Thus, a new approach should be taken into account, one that is based on honestly evaluating the impact of the emerging techniques and on the development of new strategies to overcome these menaces. This presentation is aiming to evaluate the state of the art and propose suitable solutions.

Lunch break, Coffee Time!

Template engine

This Template Engine project is a web development tool designed to generate dynamic text-based templates efficiently and consistently. It allows users to define reusable template structures and populate them with variable data to produce customized output such as HTML pages, emails, configuration files, or documentation. The engine focuses on simplicity, flexibility, and automation, reducing repetitive coding tasks while improving maintainability and development speed. By separating presentation logic from data, the project supports cleaner application architecture and easier content management.

Interactive Website for Restaurant

The Interactive Website for Restaurant project is a modern web application designed to present a restaurant’s identity, menu, and visual atmosphere in an engaging and user-friendly way. The website features a menu organized by categories, dynamic filtering for easier dish discovery, a photo gallery. Built with semantic HTML and advanced CSS techniques, the project emphasizes clean structure, responsive design, accessibility, and an attractive visual experience. Its goal is to improve online presentation, simplify menu browsing, and create a professional digital presence for the restaurant.

Functional Calculator: Implementing Basic Operations in a Web Interface

This presentation describes the development of a simple web calculator capable of performing basic arithmetic operations such as addition, subtraction, multiplication, and division. It highlights the main interface elements, the way results are entered and displayed, and the JavaScript logic used to evaluate expressions and handle delete and reset actions.

Modern Communication Networks and Security

Modern communication networks form the backbone of today’s digital society, enabling seamless data exchange across global infrastructures. As these networks evolve—driven by technologies such as 5G, cloud computing, and the Internet of Things (IoT)—they introduce increased complexity alongside expanded attack surfaces. This presentation explores the architecture and key components of contemporary communication networks, highlighting how design choices impact both performance and security. The discussion focuses on current threat landscapes, including advanced persistent threats, distributed denial-of-service (DDoS) attacks, and vulnerabilities in wireless and software-defined networking environments. It also examines the principles and practices of network security, such as encryption protocols, authentication mechanisms, intrusion detection systems, and zero-trust architectures. By bridging theoretical concepts with real-world case studies, this talk aims to provide a comprehensive understanding of how secure, resilient communication networks can be designed and maintained. The session concludes with emerging trends and future challenges, emphasizing the need for adaptive security strategies in an increasingly interconnected world.

Coffee Time and Social Networking!

WORKSHOP - Towards Data Scientist Universe Workshop

The Towards Data Scientist Universe Workshop introduces students and early-career learners to the exciting world of data science. Data science combines computer science, statistics, mathematics, and domain knowledge to solve real-world problems using data. This workshop provides a guided journey through the major areas of the data scientist’s universe, including data analysis, programming, data visualization, and ethical use of data. Participants will gain insight into how data scientists collect, clean, analyze, and interpret data to support decisions in fields such as education, business, and technology.

SpaceExploration: Cyber-resilient IoT for autonomous space habitats

Future space habitats will depend on dense IoT ecosystems that monitor astronaut health, oxygen, CO₂, temperature, humidity, and mission-critical infrastructure in real time. This session explores how cybersecurity, threat intelligence, and computational modeling can converge to protect such cyber-physical environments against data manipulation, signal interference, device compromise, SQL injection, XSS, session hijacking, and CSRF attacks. Inspired by a practical “Space Exploration” IoT monitoring platform, the talk presents a forward-looking security architecture combining zero-trust access control, tamper-evident telemetry, secure audit trails, anomaly detection, and predictive risk modeling. The goal is to show how resilient, ethical, and autonomous security mechanisms can safeguard human life and scientific data in the next generation of orbital stations, lunar bases, and deep-space missions.

Cyber criminal profiling system analysis

The escalation of advanced cyber threats needs new strategies in cyber defense, especially concerning cyber criminal profiling. This research investigates the creation and utilization of a state of the art system, crafted to harness opensource tools and frameworks. The methodology involved creating a virtualized environment utilizing VMware, constitued from an Ubuntu server hosting the Wazuh server and a Windows client running both the Wazuh agent and Sysmon. Integrating Sysmon and Atomics came to the generation of logs important for profiling cyber threats. The development of five useful Streamlit applications is the focus of this paper. These are made to examine Atomics and Wazuh logs, giving real-time insights into the strategies and methods threat organizations employ. The system is a crucial tool for cyber threat assessments since it also generates probabilities for events and permits data exports. Arrange plots are a sort of visualization that the framework gives. These plots give a picture of the cyber danger scene by interfacing methodologies to strategies and related danger bunches. More, the utilize of the jsview Chrome expansion encourages the seeing of JSON records, thus developing the lucidness and understanding of the created information. The venture serious to move forward cyber risk insights and proactive defense strategies by utilizing this improved framework. By highlighting its noteworthiness and affect, this consider has the potential to make strides cybersecurity versatility within the setting of everchanging dangers. key-words: cyber criminal profiling, wazuh, mitre attack framework, sysmon, atomics, streamlit, cyber threat intelligence, cybersecurity resilience.

From user actions to automated testing – in real time

Discover how real user interactions can instantly become automated tests. In this live demo, you will see how usage flows are captured, automatically transformed into scripts with Playwright and Locust, and organized into test suites ready to run. The application monitors performance, detects errors, and automatically generates tests to reproduce them, integrating with Grafana and Jenkins for continuous feedback. Faster. Smarter. Closer to reality.

From AI-Assisted Intrusions to Autonomous Cyber-Espionage: Kill Chain Compression, Systemic Risk and the Emerging Governance Challenge

This talk examines the transformation of cyber-espionage under the accelerating influence of artificial intelligence, focusing on the transition from AI-assisted operations to increasingly autonomous, agentic attack models. Drawing on a structured synthesis of recent threat intelligence reports, incident datasets, and documented campaigns, the study conceptualizes cyber-espionage as a systemic phenomenon shaped by temporal compression, automation of the cyber kill chain and the industrialization of social engineering. The analysis is organized along six dimensions (historical evolution, offensive capabilities, human-centric exploitation, persistence and stealth, geopolitics and attribution, and systemic risk) allowing for a comparative and evidence-based understanding of AI-driven threats. Particular attention is given to emerging autonomy thresholds, where AI systems execute a significant proportion of operational tasks, fundamentally altering attack speed, scale and strategic impact. The paper argues that these developments generate a “defenders’ paradox”, in which defensive AI capabilities fail to offset the asymmetry introduced by offensive automation. In response, it outlines a governance-oriented perspective emphasizing attribution challenges, the need for AI-specific incident regimes and a shift from prevention-based security toward systemic resilience. The study contributes to ongoing debates on cybersecurity, AI governance and national security by proposing an integrated analytical framework for understanding and mitigating AI-enabled cyber-espionage.

Lunch break, Coffee Time!

Application for Launching Attacks and Detection in Networks

This project aims to improve network testing and monitoring by creating 3 separate applications, one for launching attacks called "Casper" and 2 other being for detecting network attacks and using a dashboard where these results can be viewed, using "Casper" the most popular types of attacks on networks can be launched, either by beginner based on technical knowledge or a user with advanced knowledge, both of them will be able to use the Casper application thanks to the modern graphical interface and the easy-to-use mode specially designed to be efficient and fast. The attack detection application works together with the dashboard to display concretely and easily understandable using statistics and graphs the data resulting from the network monitor in real time.

The Cybersecurity Maze: Find your Golden Path

“Wow! Have you seen that brute-force attack? I bet I can do that. What about securing Docker? That sounds really cool. Learning cryptography would definitely impress recruiters!”. It’s easy to get lost in the excitement of cybersecurity. With so many paths — from networking and cloud to threat hunting, vulnerability management, and beyond — where do you even start? In this session, we’ll cut through the noise and help you discover the path that actually fits your skills, interests, and goals. Let’s find your Gold Path together.

Online Appointment Form with Real-Time Validation Using HTML, CSS and JavaScript

This paper presents the development of an interactive online appointment form created using HTML, CSS and JavaScript. The main purpose of the project is to collect user data and validate the information entered in real time, ensuring accuracy and preventing incomplete or incorrect submissions. The application includes essential fields such as full name, email address, phone number, personal identification number, gender, country, medical specialty, appointment date and time, password, and password confirmation. Several validation methods were implemented, including format verification for email and phone number, password strength control, matching password confirmation, and restrictions for selecting only future appointment dates. In addition, dynamic features were added, such as automatically displaying the consultation price depending on the selected medical specialty. After successful validation, the user is redirected to a confirmation page, improving usability and user experience. The project demonstrates how basic web technologies can be combined to build a practical, modern and user-friendly application suitable for real-life online scheduling systems.

Movie Database Browser Web Application

This project presents the design and implementation of a web-based Movie Database Browser application developed within the context of Dynamic Web Applications (DAW). The primary objective of the application is to provide users with an intuitive interface for searching, browsing, and exploring information about movies, including titles, release years, ratings, and short descriptions. Unlike traditional approaches that rely on external APIs such as The Movie Database (TMDB) or OMDb, this application is built using a custom backend that serves a locally stored dataset of movies. The backend is implemented using Node.js and Express, exposing RESTful endpoints that allow the frontend to retrieve and filter movie data efficiently. The dataset includes up to 250 movies, structured in JSON format, ensuring fast access and easy scalability. On the frontend, the application is developed using React, focusing on component-based architecture and responsive user interface design. Key functionalities include dynamic search, real-time filtering of movie titles, and structured display of movie information in a grid layout. The application ensures smooth interaction between client and server through asynchronous HTTP requests. The project demonstrates fundamental concepts of modern web development, including client-server communication, REST API design, state management in React, and data handling using JSON. It also highlights the advantages of building a self-contained system that does not depend on third-party services, ensuring reliability and full control over the data.

Coffee Time and Social Networking!

TaskFlow: A Modern Full-Stack Approach to Agile Project Management

This paper presents "TaskFlow", a high-performance Kanban-style application designed to streamline personal and professional project management. By leveraging a Full-Stack architecture consisting of a React-based frontend and a Node.js backend, the application ensures real-time data persistence and a fluid user experience through asynchronous state management and drag-and-drop interactions.

Image Slider / Carousel – Vanilla JavaScript

This project presents the development of an interactive Image Slider (Carousel) web application using HTML, CSS, and Vanilla JavaScript. The main objective of the application is to provide a dynamic and user-friendly way to navigate through a set of images. The slider includes features such as next and previous navigation buttons, clickable dots for direct access to slides, keyboard navigation, and an autoplay functionality that automatically changes images at a fixed interval. Additionally, the autoplay pauses when the user interacts with the slider. The application is built without external libraries, focusing on core web development concepts such as DOM manipulation, event handling, and responsive design. CSS is used to create smooth transitions, while JavaScript manages the logic and interactivity. This project demonstrates fundamental front-end development skills and can be extended in the future with features such as backend integration or dynamic content.

WORKSHOP - Geospatial Application Development: From Spatial Data to Interactive Systems

Geospatial data has become an essential component for different types of contemporary applications, including logistics, urban analytics, environmental monitoring, public services, and decision-support systems. Developing software that leverages this data presents challenges that surpass those of conventional application development, such as managing heterogeneous data formats, coordinate reference systems, spatial indexing and querying, performance optimization at scale, and designing interactive cartographic interfaces. This session provides a structured overview of geospatial application development, addressing the primary components of a typical technological stack—namely, spatial data sources and formats, spatial databases, geospatial processing libraries, and web-based visualization frameworks—along with the architectural patterns that interconnect them. Additionally, the session discusses practical considerations related to data acquisition, transformation, and integration, and concludes with an exploration of emerging trends in geospatial application engineering, including real-time spatial analytics and privacy-preserving location services.

Federated Reinforcement Learning for Antimicrobial Resistance Surveillance

Antimicrobial resistance (AMR) is a global One Health threat that demands cross-border, cross-sectoral surveillance. Centralised data aggregation, however, runs into hard limits around patient privacy, data sovereignty, and regulatory compliance. The FLAMR project (Horizon Europe MSCA Staff Exchanges, 101299901) addresses this by building a privacy-by-design Federated Reinforcement Learning (FRL) architecture connecting twelve or more laboratories across seven countries, calibrated for ESKAPEE+ pathogens. This talk presents FLAMR as a working testbed for cybersecurity research in clinical AI. The technical stack combines FRL for decentralised model training, secure multi-party computation and differential privacy at the aggregation layer, small language models (SLMs) for stakeholder interfaces, and explainable AI for clinician-facing and policy-making decision support. Classified as a high-risk AI system under Article 6 of the EU AI Act, FLAMR must address gradient leakage, model inversion, membership inference, poisoning attacks, and SLM-specific risks such as prompt injection and hallucination, while remaining auditable under GDPR and AI Act conformity requirements. The talk closes with open research questions in secure aggregation, resilient federated learning, SLM red-teaming, and AI Act conformity tooling.

Off the Rails: Geospatial Intelligence and Integrity Attacks Against Modern Logistics Infrastructure

Geographic Information Systems and the geospatial data they produce, consume, and propagate underpin an increasing share of critical infrastructure, spanning logistics and transportation, energy distribution, and emergency response. Despite this centrality, geospatial data remains underrepresented in mainstream cybersecurity threat models, and the security properties of location-bearing telemetry are often taken for granted by the systems that rely on them. This session examines geospatial security through two complementary perspectives: information leakage, in which location data reveals operational, organizational, or personal context to an adversary, and telemetry integrity, in which an adversary manipulates location-bearing signals to influence downstream decisions. The session covers representative attack and defense scenarios from critical infrastructure contexts and concludes with a discussion of open research directions in privacy-preserving spatial computation and security operations integrating geospatial data.

From bugs to contexts: G-I-A (Grounding, Instructibility, Alignment)—an ontological framework for the security of AI systems

Classical information security has built, over the last four decades, a stable conceptual vocabulary: confidentiality, integrity, availability; attack surface, vector, exploit; vulnerability, threat, risk. This vocabulary has made coordination between research, industry and regulation possible, because it has provided the categories by which an incident could be named, compared and controlled. The security of artificial intelligence systems does not yet have such a shared vocabulary. We are talking about prompt injection, jailbreak, hallucination, model poisoning, agent hijacking — operationally useful terms, but fragmented, without a categorical structure that says what is irreducible in each and how they relate to each other. This presentation proposes a minimal ontological framework for AI security, organized on three independent axes: Grounding — the property of the system to anchor its representations in verifiable references, so that semantic manipulation leaves a geometrically measurable trace; Instructibility — the property of accepting and prioritizing instructions from an explicit chain of authority, distinguishing legitimate source from adversarial injection; Alignment — the guarantee that, under the entire distribution of possible inputs, the actual behavior of the system remains the one that the organization specified. The three axes are irreducible to one another: a perfectly aligned system can be semantically unanchored, a perfectly anchored system can be poorly instructable, a perfectly instructable system can be unaligned. It is precisely this independence that transforms G-I-A from a list of concerns into a generative ontology — a framework that produces the categories through which contemporary threats become systematically thinkable. G-I-A is to AI security what C-I-A (Confidentiality-Integrity-Availability) was to classical information security. We will review the three levels of contemporary attack—code syntax, context syntax, reference semantics—and show how each level maps onto the corresponding defensive component in G-I-A. Finally, we will briefly indicate how a neuro-symbolic architecture offers a concrete way to implement Grounding, transforming semantic uncertainty from a hidden vulnerability into an auditable signal.

Quantum sensors as ultraprecise instruments to perform searches for new Physics Beyond the Standard Model (BSM) and tests of General Relativity (GR)

Quantum sensors rely on the manipulation and measurement of quantum states in atoms, photons, or solid-state defects to detect minute changes in environmental fields. They exploit phenomena like superposition and entanglement, typically using nitrogen-vacancy (NV) centers, cold atoms, or trapped ions to measure magnetic fields, gravity, acceleration, and time with extreme precision. Optical atomic clocks based on laser-cooled trapped ions and atoms have reached fractional frequency uncertainties at the 10^{-18} - 10^{-19} level, which makes them the most precise quantum sensors ever built. Current research is focused on new avenues of research aimed at further enhancing their accuracy and precision. We focus on the fundamental principles and techniques used to build optical clocks, and review the laser technologies and quantum enginnering techniques that are essential to their operation. We present remarkable progress made towards a redefinition of the SI second, scheduled in 2029. Finally, we discuss emerging applications of optical clocks and their essential role to support critical infrastructures. We also present the EU Positioning, Navigation and Timing (PNT) ecosystem initiative, which encompasses GNSS and complementary emerging alternative systems. The concept of Complementary PNT (C-PNT) is pivotal in strengthening EU autonomy, reinforcing the resilience of its economy, and advancing the EU’s global standing through technological development. Furthermore, optical clocks validate space-based tests of the General Relativity (GR) theory and offer unique platforms to perform searches for physics Beyond the Standard Model (BSM).

Securely managing workload secrets with HashiCorp Vault

This presentation shows how HashiCorp Vault can be used to centrally and securely manage secrets for users, CI/CD pipelines, Kubernetes workloads, and Azure virtual machines. We will cover integration with Microsoft Active Directory for project-based user access, GitLab JWT authentication for pipelines, Kubernetes authentication for pods, and Azure authentication with Vault Agent for VMs. The focus is on replacing scattered static secrets with identity-based access, short-lived credentials, clear segregation, and centralized auditing.

Lunch break, Coffee Time!

Architecture and Development of a Secure Full-Stack Web Platform for Enterprise IT Services Management

This paper presents the design and development of Nova IT Solutions, a full-stack web platform for an IT services company specializing in network infrastructure, cybersecurity, and cloud computing. The platform integrates a modern, animated front-end interface with a functional Node.js back-end, SQLite database, and a secure session-based authentication system. A key contribution of this project is the implementation of a Security Proof of Concept, which demonstrates practical application of web security principles including password hashing with bcrypt, protected API routes, session management, and input validation. The interactive IT Equipment Generator recommends real-world hardware configurations from vendors such as Cisco, Fortinet, and Palo Alto Networks based on company profile. The administrative dashboard provides authenticated access to contact message management. This project demonstrates practical competencies in full-stack development, RESTful API design, secure authentication architecture and modern UI/UX engineering.

WORKSHOP - The Cybersecurity Maze: Not all paths are yours

“Wow! Have you seen that brute-force attack? I bet I can do that. What about securing Docker? That sounds really cool. Learning cryptography would definitely impress recruiters!”. It’s easy to get lost in the excitement of cybersecurity. With so many paths — from networking and cloud to threat hunting, vulnerability management, and beyond — where do you even start? In this session, we’ll cut through the noise and help you discover the path that actually fits your skills, interests, and goals. Let’s find your Gold Path together.

Design and Analysis of a Distributed Microservices-Based System with Go and MongoDB

This paper presents the design and implementation of a distributed system based on a microservices architecture, developed using the Go programming language and MongoDB as a NoSQL persistence layer. The system is composed of three independent services (Authentication, Product Management, and Order Processing) each responsible for a distinct business domain and supported by its own database. The implementation highlights key architectural principles, including service decoupling, the database-per-service pattern, RESTful inter-service communication, and stateless authentication using JSON Web Tokens (JWT). Additionally, the system is designed with containerization in mind, enabling scalable and modular deployment using Docker. The project provides a practical exploration of fundamental distributed system challenges, such as scalability, fault isolation, and data consistency, while demonstrating how modern backend systems can be structured for flexibility, maintainability, and extensibility.

Real-time Chat Backend

This presentation explores the design and implementation of a scalable real-time chat application built using Node.js, Socket.IO, and a relational database backend. The system is architected with a clear separation of concerns, where the HTTP server, session middleware, and WebSocket communication layer are initialized independently to improve modularity, maintainability, and scalability. At the core of the architecture, Socket.IO enables real-time bidirectional communication between clients and the server, while session-based authentication ensures that only authorized users can establish and maintain connections. Each socket is securely linked to authenticated user data, preventing unauthorized access and enforcing communication integrity. The application supports dynamic room creation and management, allowing users to join, leave, and switch between chat rooms seamlessly. Messages are validated, persisted in a database, and broadcast only to relevant participants using efficient room-based event emission. This ensures both real-time delivery and long-term data persistence. Additionally, the system retrieves chat history on demand, enabling a continuous user experience across sessions. A modular handler structure is used to separate socket events into independent components, improving code organization and scalability for future feature expansion.

Cyber-Fusion Horizons: Securing the experimental science of tomorrow

The next generation of scientific infrastructure will merge plasma physics, autonomous sensing, computational modeling, and secure data ecosystems. Inspired by the transition from cold plasma experiments to controlled fusion, this lecture explores how cybersecurity, threat intelligence, IoT, and computational modeling can protect the future of high-impact scientific research. The session focuses on secure acquisition, storage, transmission, and validation of experimental data generated by plasma diagnostics, material characterization systems, and fusion-oriented laboratory platforms. It introduces IoT-aware monitoring of experimental environments, and AI-assisted threat intelligence for detecting anomalies, tampering, and unauthorized access. By combining plasma–material simulations, Python/MATLAB-based analysis, cryptographic protection, and zero-trust principles, the topic frames fusion research as a cyber-physical frontier where trusted computation and secure scientific data become as critical as the experiment itself.

Aruba vs Cisco? Why Not Both. Lessons from a Real Hybrid Enterprise Network

Modern enterprise networks rarely belong to a single vendor. In our environment, Aruba and Cisco coexist by design: Aruba switches and access points provide the foundation for wired and wireless connectivity, managed on‑prem through Aruba AirWave and Aruba VMC, while Cisco delivers the security and identity layer through Cisco ISE, Cisco firewalls, and Cisco FMC for centralized policy management.

Constantin Carabella National College

ITLevel Partner

National Computer College Tudor Vianu

ITLevel Partner

CTF (Capture-the-Flag) - by Razvan Mihai Iacob, ITLevel

Razvan Mihai Iacob is a Romanian ethical hacker, security researcher, and software developer with a strong focus on low-level systems and cybersecurity. He is currently studying at the Tudor Vianu National College of Computer Science, where he has built a solid foundation in computer science and competitive programming. Through his personal platform, he showcases hands-on research in areas such as binary exploitation, kernel vulnerabilities, and reverse engineering. His technical work includes detailed write-ups on topics like Linux kernel use-after-free vulnerabilities, heap exploitation techniques, and advanced attack chains (e.g., arbitrary write to remote code execution in glibc internals), reflecting a deep engagement with modern offensive security challenges. Razvan has achieved notable competitive success in cybersecurity and informatics. Among his recent results, he qualified for the finals of the Romanian Cyber Security Challenge (ROCSC) 2026, reached the national stage of both the Romanian AI Olympiad and the Applied Informatics Olympiad (ACADNET), and earned a bronze medal at the national AI Olympiad. He also secured 1st place at Unbreakable Romania Teams 2026. In previous years, he ranked 1st in his grade and 3rd overall at the National Cybersecurity Olympiad of Romania (2025), placed 5th at the Romania Cyber Security Challenge finals, and achieved a top-6 finish at DefCamp CTF finals with his team. Through continuous participation in CTF competitions and the publication of in-depth technical analyses, Razvan demonstrates both practical expertise and a research-driven mindset. His work reflects a commitment to advancing knowledge in cybersecurity and contributing to the broader technical community. More details here.

Summer School Closing Remarks