Join Embedl at EDGE AI Milan 2025: Where Innovation Meets Impact
From July 2 to July 4, 2025, the global edge AI community will gather in Milan, Italy, for EDGE AI Milan 2025—a premier event at the intersection of advanced research, real-world applications, and next-generation AI deployment. Embedl is proud to participate in this future-focused conference, which will bring innovators, researchers, and industry leaders together to explore the transformative power of edge AI across sectors.
Why Edge AI, Why Now?
As AI evolves from cloud-centric to embedded directly in devices, robotics, and autonomous systems, the demand for edge computing is growing rapidly. Edge AI presents a unique opportunity to deliver intelligent capabilities closer to where data is generated—reducing latency, conserving bandwidth, and enabling real-time decision-making in power- and resource-constrained environments.
From tinyML and neuromorphic hardware to generative AI and intelligent sensing, EDGE AI Milan 2025 promises a deep dive into the technologies actively reshaping domains such as healthcare, manufacturing, infrastructure, retail, and more.
Catch Jonna Matthiesen on the Mainstage
Embedl's own Jonna Matthiesen, a Deep Learning Researcher, will take the stage on July 2 at 10:00 AM to present a session titled:
"The Optimization Trap in Edge AI – Insights from Optimizing LLMs and Vision Models Across Embedded Platforms"
What the Session Covers
This talk challenges the assumption that smaller models are inherently faster and better suited for edge environments. Drawing from large-scale benchmarks of vision backbones and small language models (SLMs) on GPUs, CPUs, and DSPs, Jonna reveals a more nuanced picture: performance is not strictly tied to model size but is deeply influenced by the match between model architecture and hardware characteristics.
Attendees will learn how to:
- Avoid costly mismatches between model architecture and target hardware
- Select accelerators based on throughput metrics specific to model types
- Design hardware-aware optimization pipelines that are both task- and device-conscious
This session offers actionable insights into deploying reliable, high-performance vision and language models under strict latency and power constraints—an essential skill set for anyone working in embedded AI.
Looking Ahead
As Edge AI continues to push boundaries, events like EDGE AI Milan serve as vital platforms for knowledge sharing, collaboration, and community building. Embedl invites researchers, engineers, product leaders, and decision-makers to join us in Milan to explore what's next in edge intelligence.
We hope to see you there; visit our demo booth, attend Jonna's session, or schedule a meeting here!
Follow us on LinkedIn for live updates during the event!