Meet us at GAIA!
Join us as Jonna Matthiesen, Deep Learning Researcher at Embedl, takes the stage with her talk:
"Evaluating Retrieval-Augmented Generation Systems: Challenges and Practices"
Room: H1/H2
Time: 3:00 PM
Generative AI for Resource-Constrained Devices
Generative AI—especially Large Language Models (LLMs) and Transformers—is revolutionizing real-time situational awareness, autonomous decision-making, and data processing. However, deploying these models via traditional cloud-based solutions presents connectivity challenges, cybersecurity risks, and high operational costs.
This talk explores a systematic approach to deploying advanced generative AI models directly on edge devices—such as autonomous vehicles, drones, and IoT systems—to ensure autonomy, security, and real-time performance.
Key Topics:
✅ Evaluating SoCs for running generative AI workloads on low-power hardware
✅ Model compression techniques (quantization, pruning, knowledge distillation)
✅ Trade-offs between model fidelity and real-world feasibility
✅ Case study: Running SLMs (e.g., Meta’s Llama 3.2) on Qualcomm Snapdragon SoCs
Learn how to:
✔️ Assess hardware accelerators for low-latency, high-throughput inference
✔️ Optimize AI models without sacrificing performance
✔️ Future-proof AI applications using emerging trends in SLMs
About Jonna Matthiesen
Jonna is a Deep Learning Researcher specializing in AI optimization for defense, automotive, and IoT applications. With expertise in hardware-aware neural architecture search, model compression, and inference optimization, she focuses on making LLMs deployable in embedded systems and edge devices.
She holds a B.Sc. in Mathematics from Kiel University, Germany, and an M.Sc. in Applied Data Science from Gothenburg University, Sweden. Since 2023, she has been part of Embedl, a company dedicated to efficient deep learning solutions for automotive, defense, and IoT.
Don't miss this deep dive into next-generation AI deployment!
Get your ticket here: https://www.gaia.fish/2025#Home
See you at GAIA!