Breaking Barriers Deploying AI at the Edge – Conversations in the Cloud – Episode 275

Daniel Chang, CEO of ENERZAi, joins host Jake Smith to discuss ENERZAi’s vision of delivering the best AI experience on everything for everyone, and how they are doing this by overcoming the constraints that edge devices have through AI models. He highlights how ENERZAi’s Automated Model Compression Optimization Toolkit enables AI models to maintain high accuracy, while minimizing latency, size, power consumption for successful Edge deployment, as proven in their recent collaboration with Intel. Daniel further illuminates how their collaboration through the Intel AI Builders program optimized their state-of-the-art 3D hand pose estimation model for Intel Xeon processors achieved incredible performance results. Utilizing the Intel OpenVINO toolkit to improve the latency and inference times while not compromising the model’s accuracy. This optimization project also helped pave the way for customers aiming to use ENERZAi’s 3D hand pose estimation solution for their driver monitoring systems, AR/VR systems or other systems on Intel processors in an incredibly performant way. Jake and Daniel also chat about a customer use case where ENERZAi was able to help a SAAS customer migrate their solution from expensive GPU instances to more cost efficient Intel CPU instances while preserving the model accuracy. Lastly they both dive into discussing the future of AI and how it can solve the constraints that edge devices face to truly enable AI to be deployed everywhere in the world.

For more information, visit:
enerzai.com

Follow Jake on Twitter at:
twitter.com/jakesmithintel