Sylvester Kaczmarek

Architecting Secure Hybrid Intelligence for Safety-Critical Space Missions

This image from NASA's Artemis I mission symbolizes the frontier for my work. Hardware can reach new worlds, but trustworthy AI is key to operating there safely. My mission is to architect that intelligence. I build the secure, verifiable AI foundation future autonomous systems need as humanity expands into space.

My general fields are advanced artificial intelligence and cybersecurity for safety-critical cyber-physical systems, all unified by a focus on first-principles safety and stability. I specialize in architecting and leading the development of secure-by-design and interpretable hybrid AI/ML systems, leveraging high-performance edge and disruptive computing paradigms to enable new mission concepts and improve mission trade-offs. These systems combine a centralized AI core with distributed edge AI, using in-memory computing (IMC) architectures, general-purpose processors for system management and control, including radiation-hardened FPGAs, RISC-V, and ARM-based SoC cores, HW/SW co-design, digital signal processing (DSP), and spatial data processing.

This involves providing rigorous proof of safety, computational efficiency, and performance for safety-critical autonomous space systems, directly mitigating risks to life and infrastructure, laying the groundwork for increasingly complex and ambitious off-world operations.

I focus on incorporating local on-board software and data processing, which is orchestrated by multi-purpose system-on-chip (SoC) management cores or microcontrollers, and decentralized node intelligence, using distributed robust planning algorithms to coordinate multiple autonomous vehicles navigating and operating in dynamic uncertain environments, combined with numerical simulation in state-of-the-art supercomputers.

This approach, encompassing both the cyber-physical system as a whole and its constituent mechatronic elements, embodies a proven architectural paradigm for complex safety-critical systems: centralized command for strategic coherence and safety assurance, paired with decentralized execution for adaptive resilience in dynamic environments. This architecture is essential for the future of autonomous operations in space, enabling increasingly complex and ambitious robotic missions far beyond Earth.

My work in this domain is actively supported by key national security and space innovation bodies including the UK Space Agency, the UK's National Cyber Security Centre (NCSC, part of GCHQ), and the U.S. Space Force's SDA TAP Lab. Furthermore, my broader contributions to advancing Artificial Intelligence for Space Exploration and Humanity have been recognized with an Award of Merit by the European Space Agency (ESA) and Frontier Development Lab (FDL) Europe. My expertise is also sought at the highest levels of policy, contributing expert analysis and strategic recommendations to the UK Parliament's House of Lords Select Committee on UK Engagement with Space on AI, autonomy, cybersecurity, and governance. This expertise is strengthened by specialised trainings and certifications from organisations such as NATO, The Alan Turing Institute, CISA, FEMA, IAEA, and the Cloud Security Alliance.

My efforts significantly advance the safety, efficiency, and sustainability of space-based robotic intelligence systems and cislunar mobility, leveraging modern AI (deep reinforcement learning, multimodal large language models (LLMs), graph neural networks (GNNs), self-supervised learning, event-based vision, spiking neural networks (SNNs), and agentic systems), while pioneering the foundational frameworks required to formally verify their emergent behaviours. This synthesis of cutting-edge AI with verifiable control, using neuroscience-inspired architectures, is essential to safeguarding critical infrastructure through work with organisations including NASA, ESA, UKSA, NATO, and The Alan Turing Institute, while advancing OS’s robotic intelligence programme and setting new standards for autonomous space operations. This work is paving the way for a future where autonomous systems can operate reliably and safely far beyond Earth's immediate vicinity.

In parallel to these technical pursuits, my approach is deeply aligned with the principles of machine alignment, focusing on embedding human values like fairness and sustainability into AI. This involves developing systems grounded in verifiable, axiomatic mechanisms derived from first principles through the enforcement of hard operational constraints at the system's dynamic core. This foundation enables AI that is transparent in its reasoning, amenable to expert knowledge integration, and ultimately capable of being controlled effectively, ensuring that these systems benefit humanity as they evolve.

I emphasise the integration of ethical frameworks and stakeholder engagement in the AI development process to foster systems that are not only fair and sustainable but also robust against ethical dilemmas. Addressing the challenge highlighted in Geoffrey Hinton's Nobel address, where he warned that advanced AI systems "are no longer science fiction" and that "we urgently need research on how to prevent these new beings from wanting to take control", these ethical considerations are absolutely critical. They are essential to ensuring a future of harmonious coexistence between humanity and superintelligent machines, a future that extends throughout our solar system.

Credit: NASA / William Anders, Apollo 8

The "Earthrise" photograph from Apollo 8, often called the most influential environmental photo ever taken. This view of our fragile home is the ultimate "why" for my work. It is a constant reminder that the intelligent systems we build must be embedded with the wisdom and responsibility this perspective demands.

Driving questions like "How to build robust and trustworthy AI for complex autonomous systems in space?", and the ultimate challenge of "When does consciousness emerge from AI's perceptual abilities?", capture my pursuit in AI safety and security, autonomous systems, and space exploration. Through rigorous interdisciplinary research and collaboration, I am building solutions to these critical challenges, striving to harness the intricacies of AI for ethically aligned and beneficial systems that evolve alongside humanity. These solutions are not just for today's missions; they are the building blocks for a future where humanity thrives among the stars.

This entire endeavour answers the simple, profound challenge articulated by Alan Turing in 1947: "a machine that can learn from experience." In essence, my work advances the foundational pursuit defined in the 1955 Dartmouth Conference proposal: "...the science and engineering of making intelligent machines," now applied to the new, critical domain of AI safety and security, enabling humanity's expansion beyond Earth.

The mission is just beginning.