EDGE AI Neuromorphic WG Oct 29, 2026

Bi-weekly meeting

Present:

NameAffiliationEmail
Petrut BogdanInnaterapetrut.bogdan@innatera.com
Pete BernardEDGE AI Foundationpete@edgeaifoundation.org
Jonathan TapsonBrainChipjtapson@brainchip.com
Rahim RahmaniStockholm University (DSV)rahim@dsv.su.se

Attachments EDGE AI Neuromorphic WG EDGE AI Neuromorphic WG kick off meeting – 2025/09/29 08:56 CST – Notes by Gemini

Summary

Taipei event updates and neuromorphic hackathon feedback via Munich workshop planning and benchmark validation

Taipei Trip And Munich Workshops
Expanded Taipei event success despite typhoons prompting plans for Munich hands-on neuromorphic workshops.

Neurobench Results And Virtual Panels
Validation proposals for Neurobench combined with tentative plans for multi-topic virtual panel sessions.

Infrastructure White Paper And Scheduling
White paper skeleton drafting initiated alongside proposed adjustments to meeting time zones.

Decisions

Aligned

  • Topic-focused holiday neuromorphic panel sessions The group aligned on organizing an early December neuromorphic event featuring multiple topic-focused panels back-to-back.

Next steps

  • unticked[Pete Bernard] Draft Job Description: Write a job description for a dedicated benchmark project manager. Share this draft with the team to gather feedback.
  • unticked[Petrut Bogdan] Coordinate Fireside Chat: Send an email to prospective participants to verify availability for a holiday neuromorphic fireside chat. Determine whether the event requires a single panel or multiple topic-specific sessions based on responses.
  • unticked[Jonathan Tapson] Draft White Paper: Outline the skeleton structure for a white paper summarizing the previous live stream presentations. Coordinate with contributors to finalize the content sections.
  • unticked[Vijay] Lead Infrastructure Effort: Manage the coordination of specialized sessions dedicated to neuromorphic infrastructure.
  • unticked[Jonathan Tapson] Verify Meeting Access: Confirm whether invitees can access the meeting notes document. Report findings to the team.
  • unticked[The group] Prune Distribution List: Clean up the email distribution list for the meeting series to maintain an accurate invitee list.
  • unticked[The group] Plan Workshop: Organize a hands-on neuromorphic workshop for the upcoming Munich event.

Details

Taipei Trip and Event Update: Pete Bernard shared an update on their third year attending the event in Taipei, noting a significant increase in size and activity with two simultaneous workshop rooms, a main stage, and demo areas (00:00:01). The event included a startup pitch contest featuring ten startups, where Simit served as a judge. Proceedings on day two were disrupted by a typhoon and a government-mandated shutdown (typhoon holiday), which also delayed Pete Bernard’s return flight from Saturday to Monday (00:01:24). Despite weather disruptions, the trip was deemed very positive, resulting in meetings with several Taiwan domestic technology companies and startups such as MediaTek, Realtek, Advantech, and various robotics companies (00:02:22).


Neuromorphic Presence and Future Workshops in Munich: Regarding neuromorphic representation at the Taipei event, Pete Bernard noted that BrainChip was present, but there were no standalone hands-on workshops like the usual Dell, Intel, or Arduino sessions (00:03:17). Pete Bernard suggested organizing a hands-on neuromorphic workshop for the upcoming event in Munich (00:04:21).


Neuromorphic Hackathon Feedback: Jonathan Tapson and Petrut Bogdan discussed the recent Neuromorphic Hackathon. Jonathan Tapson reported that Alf was significantly involved and that online enthusiasm was high (00:04:21). Petrut Bogdan attended for a day and a half, observing high engagement from students, three live demos using Pulsar, and one data set-driven pulsar-in-the-loop demo (00:05:21). Petrut Bogdan suggested that the AGI Foundation could co-organize or support future hackathons by serving as a user base and problem-space provider (00:06:29).


Neurobench Results Validation and Benchmarking PM Proposal: Pete Bernard raised a question from Sumit regarding whether an organization like the foundation should help independently validate and aggregate results for the Neurobench challenge (00:09:15). Petrut Bogdan supported the idea of a central aggregation and independent validation space similar to MLCommons (MLPerf) to avoid disjointed academic reports and ensure impartial measurement checks (00:10:37). Pete Bernard proposed budgeting for a dedicated benchmark product manager next year to drive the aggregation and validation process, starting with Neurobench, and agreed to draft a job description for feedback (00:13:12) (00:15:30).


Fireside Chat and Virtual Panel Organization: Petrut Bogdan proposed organizing an online virtual fireside chat or panel discussion in early December (before December 15th) to follow up on the recent live stream. Pete Bernard agreed to support the event (00:08:02) (00:16:24). While Petrut Bogdan initially envisioned a single one-hour session with about five people and a moderator, Pete Bernard, Rahim Rahmani, and Jonathan Tapson discussed splitting the event into multiple topic-focused back-to-back panel sessions (such as hardware, algorithms, and student career development) to better engage audiences and accommodate different interest areas (00:18:28) (00:22:07). Petrut Bogdan agreed to send an email around with a tentative date to gauge availability and determine whether to run multiple panels (00:24:33).


Infrastructure Discussion and White Paper Collaboration: Petrut Bogdan noted that a separate discussion on neuromorphic infrastructure led by Vijay would need to be driven further (00:24:33). Additionally, Petrut Bogdan requested assistance in coordinating and drafting a white paper or perspective article based on the live stream presentations. Jonathan Tapson agreed to help with drafting the document skeleton and inviting contributors (00:25:36).


Meeting Logistics and Scheduling: Pete Bernard followed up on meeting notes distribution via Google Docs, which Petrut Bogdan confirmed receiving for neuromorphic topics. Pete Bernard and Jonathan Tapson discussed pruning the large invite list and addressing scheduling conflicts caused by time zone differences, noting the current heavy Eurocentric bias (00:26:50). Pete Bernard and Jonathan Tapson proposed alternating meeting times between morning and evening slots every few weeks to better accommodate participants in Australia, Hong Kong, India, and other non-European time zones (00:27:51).