2024-10-15

Owned by Tom Qin
Last updated: Oct 17, 2024 by AKRAM SHERIFF
3 min read

Agenda

  • target use cases for OPEA and InfiniEdge AI collaboration @Wilson Wang @Shunli, 15 mins

 

Participants

Please list your name here if you attend this meeting. @Tina Tsou @Wilson Wang @Victor Lu @Qi Tang @Haruhisa Fukano @Kris He @Borui Li (李博睿) @Moshe Shadmon Jinpu Cao, Li Shun, @AKRAM SHERIFF

 

Recording

https://bytedance.us.larkoffice.com/minutes/obusct2n7r1qg7r6r4g24777?from=from_copylink

Summary

The meeting discussed how Opia and Infinite Edge AI can work together on the Micro Service Framework, cscd, and the AI Agent platform, the main contents included:

  • Architecture of Infinite Edge AI: Wilson Wang introduced the general architecture of the infinite edge aspire project, including the agent runtime, model, and agent tools management.

  • Agent Build Platform: It is for general UE agent and supports other agents like long flow.

  • Backend Models: The back end supports open source models, ByteDance models, and user's own models.

  • OPI and Infinite Edge AI Collaboration: They need to check how to make a compatible version to support both APIs or use part of OPR.

  • Use Cases: Victor and Mao Share introduced the use cases of manufacturing and smart cities.

Chapters

01:05 Discussion on Collaboration between Opia and Infinite Edge AI on Micro Service Framework, cscd and AI Agent Platform

This section focuses on the discussion between Opia and Infinite Edge AI regarding their potential collaboration on the Micro Service Framework, cscd, and the AI Agent platform. Wilson Wang presents the general architecture of the Infinite Edge Aspire project, including components for AI agents. They discuss aspects like the Agent Build platform, different agent frameworks, and backend support for various models. They also plan to check API mappings and set up a next call with engineering teams for more detailed discussions.

18:14 Jingpu's Pavement Maintenance Project and Related Use Case Discussions

This section begins with Jinpu, a PhD student from civil and environmental engineering, introducing his work on developing a method to check pavement distress over time using cameras on vehicles and algorithms. He wants to discuss with public works departments for comparison. Then, Tina Tsou tries to have someone introduce a relevant use case in the overview document, but there are some misunderstandings and complications in this process.

25:05 Edge Lake: Decentralized Data Management at the Edge for Various Industries

This section is about an presentation of a system related to edge computing. The presenter shows green dots which are edge instances worldwide. Data from sources like wastewater facilities goes to nodes at the edge. It can be interacted with as if centralized using virtualization. AI is also being enabled on this setup and progress is being made. There are also discussions about handling data from various platforms, data privacy, security and the benefit of not using cloud services.

33:38 Project Updates and Demo in AI - related Work Streams

This section mainly discusses work plans and updates. Tina asks Jinpu to write up the problem statement. There are upcoming events like the IBM conference that may affect participation. Wilson skips the overview update due to POC work. They discuss various workstreams including Geo distribute TikTok for agents, database and algorithms. Wilson also gives a demo related to the spear platform and they discuss aspects like API support for agent platforms.

46:56 Updates on various work streams including federated learning, edge gateway, robotics and TikTok at a meeting

This section is an update meeting. For federated learning, they are working with a company and will show a demo in a few weeks for use cases like smart city. Work on a lightweight protocol will be completed in 2 - 3 weeks. Updates on robotics include potential Stanford collaboration. A meetup may be rescheduled to next Friday. Luma side may need an update too.