Hunyue Yau from HYR, LLC (http://www.hy-research.com/) is an embeded Linux consultant, software and hardware developer and enthusiast with over a quarter century involvement in Linux touching on multiple Linux architectures including x86 and ARM. Areas of interest include almost any sensor/sensing technology, machine learning/AI low power and small footprint for embedded/mobility with focus on hardware/low level Linux infrastructure. Prior works include developement of one of the first Linux appliances and Android porting to new hardware. Recent work range from bringup, integration, and debugging of ARM systems with custom, Yocto/OpenEmbedded, and Android userlands. He is the author of Learning BeagleBone and has presented at venues including ESC, ELC, SCaLe, and MontaVista Vision on numerous topics from power management, system sizze reduction, bringup techniques and mobile related technologies.

Presentations

23x

Leveraging LLMs on embedded Devices

Leveraging LLMs (Large Langage Models)/machine learning in an embedded environment can be riddled with surprises and challenges due differences on embedded devices and expectations. This session will look at challenges encountered by an embedded developer evaluating LLMs on an embedded Linux device along with trade offs in trying to fit an open LLM on an embedded device. The challenges will be illustrated with data from different attempts attempts on embedded Linux. Combination of both hardware and software will be looked at to address the challenges.

See Presentation