Quantum Computing for AI
This presentation will discuss the potential of quantum computers to process high-dimensional data spaces, accelerate optimization problems, and offer new representation methods for machine learning models. At the same time, it will emphasize that current quantum hardware remains limited and that quantum advantage for artificial intelligence is still at the research stage in many applications. Quantum computing will therefore be presented not as a near-term replacement for classical artificial intelligence, but as a complementary technology for specific classes of problems.
Tracing the Agent's Mind
Production AI agents make thousands of tool-calling decisions daily, yet observability stops at the model boundary. OpenTelemetry's GenAI semantic conventions capture token counts and latencies—what the LLM processed—but not why an agent selected a specific tool. Research (McKenzie et al., 2023) demonstrates inverse scaling: more capable models exhibit unpredictable tool selection patterns. This gap leaves engineers guessing during critical production failures.
From Chatbots to Engineers: Trustworthy, Sovereign AI Agents for Engineering
Foundation-model agents promise to reshape engineering: from requirements to design to verification, but engineering is not open-domain chat. Along the engineering V, every specification must trace to a check, and errors are costly, safety-critical, and regulated. In this talk, we discuss how engineering-grade agents can benefit from a neuro-symbolic approach: foundation models supply flexibility, while knowledge graphs and ontologies supply the domain grounding, constraints, and machine-checkable semantics that make outputs auditable and traceable across the V. The talk further makes the case for sovereignty: compact, self-hosted models grounded in knowledge graphs that rival larger closed systems while keeping data and IP in-house.
Human Centered Computer Vision Systems & Applications
Facial image analysis has long been a central problem in computer vision, aiming to automatically extract rich information about individuals, such as identity, affect, behavior, and other soft biometric traits. Beyond its traditional role in biometrics, face perception has become a key enabler of human-centered computing, supporting natural and intuitive human–computer interaction in applications ranging from security and smart environments to social robotics and multimedia understanding.
