FDL (loading) Program

Abstract

The role of modeling is undergoing a fundamental shift. Traditionally, models have been used to document, specify, and communicate complex systems. However, in an era increasingly shaped by artificial intelligence, distributed systems, and rapidly evolving requirements, this is no longer sufficient.

This keynote argues that modeling must evolve from a descriptive discipline into an execution-driven capability. Instead of creating models that merely represent systems, we need models that actively shape decisions, enable automation, and provide structure for AI-driven reasoning.

Drawing on real-world experience from enterprise architecture, model-based systems engineering (MBSE), and large-scale industrial deployments, this talk explores:

• Why traditional modeling approaches are losing relevance in modern engineering environments
• How AI changes the expectations towards structure, semantics, and traceability
• What it takes to turn models into actionable assets across the lifecycle
• How modeling can bridge the gap between engineering rigor and business impact

The keynote challenges the audience to rethink modeling not as an artifact, but as a strategic capability - one that enables organizations to manage complexity, increase resilience, and unlock the full potential of AI-assisted engineering.

Speaker Bio

Peter Lieber is a software entrepreneur and business developer with over 20 years of experience in model-based engineering and enterprise architecture. He plays a key role in building and scaling the European ecosystem around Sparx Systems, one of the world's leading platforms for modeling and architecture. His work focuses on bridging the gap between engineering methodologies and real-world business impact, particularly in domains such as MBSE, enterprise architecture, and AI-supported system design.

In addition to his work in the software industry, he is also active as an entrepreneur in sustainable retail and serves as president of the Austrian Gewerbeverein, representing SMEs and entrepreneurial interests. Peter is known for combining deep technical understanding with a pragmatic, business-driven perspective on modeling and digital transformation.

Abstract

With formal proofs of its correctness and security, seL4 provides the strongest assurance and isolation between untrusted and trusted components running in critical software systems, demonstrably preventing cyber attacks from endangering critical functionality and from propagating further. seL4 is being adopted in a number of sectors, from automotive to IoT, Defense, and more. The open-source seL4 Foundation provides the underlying support for its growing ecosystem of developers, contributors and adopters, and Proofcraft keeps pushing the roadmap of seL4's verification to make its formal proofs apply to more platforms, more features, and more properties. The talk will give an overview of seL4's journey and roadmap ahead.

Speaker Bio

June Andronick is CEO and co-founder of Proofcraft, a company providing commercial support for software verification in general and the seL4 microkernel verification in particular. She is also CEO of the seL4 Foundation, supporting the technology and community of seL4 users, adopters and developers. June has extensive leadership experience towards making the vision of verified software a reality in mainstream critical software. She was previously leading the Trustworthy Systems group, and contributed to the original seL4 verification. She holds a PhD in Computer Science from the University of Paris-Sud, France.

Abstract

The rapid growth of system complexity - driven by AI workloads, heterogeneity, and aggressive technology scaling - is challenging traditional approaches to design and specification. While design and specification languages remain essential, productivity is increasingly limited by the ability to explore large design spaces, connect abstraction levels, and close design and verification loops efficiently.

This keynote presents an industrial perspective on how AI-powered and agentic EDA technologies are reshaping design flows for complex hardware and cyber-physical systems. By tightly integrating generative and agentic AI with existing languages, models, and signoff-grade tools, engineers can move from point automation to closed-loop, tool-in-the-loop workflows spanning specification, RTL design, verification, optimization, and debug.

Through concrete examples, the talk illustrates how AI agents augment - not replace - formal methods and design languages, enabling scalable, trustworthy co-design and pointing toward future evolutions of design and specification ecosystems.

Speaker Bio

Gianvito Lorusso is Senior Director of Project and Program Management at Synopsys, where he leads global, multi-site programs in the semiconductor and EDA domain, coordinating hundreds of engineers across advanced technology nodes. With more than 25 years of experience spanning ASIC design, verification, physical implementation, and large-scale SoC programs, he has worked across the full spectrum from hands-on engineering to strategic leadership.

Throughout his career at Synopsys, Cadence, and semiconductor design service companies, he has managed complex industrial projects for leading hyperscalers and semiconductor companies, including high-performance, automotive-grade, and AI-driven systems. His current focus includes the integration of AI and agentic technologies into EDA workflows, and the evolution of design and verification methodologies for increasingly complex systems.

Gianvito holds an M.Sc. in Electronic Engineering and is a PMP-certified professional.

Panelists

Abstract

Moving computation from the cloud to the device is reshaping how embedded, cyber-physical, and IoT systems are specified, designed, and verified. On-device workloads are pushing heterogeneous SoCs, often built around RISC-V platforms, toward conflicting goals: energy efficiency, real-time guarantees, fault tolerance, and secure edge AI. This special session addresses these goals by combining efficient hardware/software co-design, formal modeling, simulation infrastructure, and verification techniques, treating trustworthiness as a system-level property achieved through the co-design of architectures and methods.

Topics of interest include, but are not limited to:

• Open and configurable hardware platforms
• Hardware/software co-design for on-device AI
• Approximate and energy-efficient computing
• Reliability and dependability
• Hardware security and secure edge AI
• Languages, modeling, and verification
• Simulation, digital twins, and cyber-physical systems

Abstract

The 80's have seen the substitution of schematics by languages to describe digital circuits. Since then, standardization has gone its way, and current tools allow very efficient digital circuits to be produced from textual descriptions. Nevertheless, with growing circuit complexity, higher abstractions have been searched for while keeping full control over the produced circuits. Using languages at first designed for software, many hacks, more or less formalized, have been developed as code generators. With time, the need for a more formalized approach has emerged, as has the Hardware Construction Language community. This Special Session aims at gathering language designers and power users of these languages to discuss the current state and possible evolutions.