Edge AI Virtual Agents

Project Technical Lead:  

Project Committers detail:

Committer

Committer

Company

 Committer Contact Info

Committer Bio

Committer Picture

Self Nominate for PTL (Y/N)

Qi TangIndividual developerqitang2023@gmail.comPh.D., Hardware Engineer at Google, Ex-Meta, Connectivity Research, ETSI MEC Delegate

Qi Tang

Y
Wei Wu
willwuwork@gmail.comLLM researcher at Amazon



Yi HanIndividual developeryihan7206@gmail.comDeveloper, Finance Manager at Achieve, M.Sc University of Southern California; B.S Peking University



Sharu JiangIndividual developerjsrshark110@gmail.comSoftware Engineer at Google, Tech LeadTech lead at Google



Presentation

a video presentation: https://www.youtube.com/watch?v=r2GfqvbA0hk&t=360s

Use Case Details

AttributesDescriptionInformational
Type

New Blueprint for AI Services  at the Edge


Blueprint Family - Proposed Name

AI Edge Blueprint Family


Use Case

Running Real-Time Generative AI Models at the Edge


Blueprint proposed Name

Serverless


Initial POD Cost (capex)

Leverage Unicycle POD - less than $150k


Scale & Type

Up to 7 servers

  x86/ARM server or deep edge class

  GPU Accelerators

  Memory Cache


Applications

Edge Servers Hosting Real-Time Generative AI Services, e.g. Speech-to-Speech Translation, LLM-based Virtual Assistant Services, Text-to-Image/Video Content Generation Services, etc.


Power Restrictions

Less than 10Kw


Infrastructure orchestration

Kubeless

Docker 1.13.1 or above and K8s 1.10.2 or above- Container Orchestration

OS - Ubuntu 16.x

Under Cloud Orchestration - Airship v1.0


SDN

OVS


Workload Type

Containers


Additional Details

Video Streaming: gRPC, 

Pytorch, Tensorflow

Non-Sql Database: MongodB