diff --git a/tags/tag-workloads-foundation/subprojects/batch/meeting-notes/2026-06-30.md b/tags/tag-workloads-foundation/subprojects/batch/meeting-notes/2026-06-30.md
new file mode 100644
index 000000000..330e1e5c5
--- /dev/null
+++ b/tags/tag-workloads-foundation/subprojects/batch/meeting-notes/2026-06-30.md
@@ -0,0 +1,50 @@
+# Meeting Notes — Jun 30, 2026
+
+📽️ [Recording](https://www.youtube.com/watch?v=5n7vdaZyoho)
+🤖 [AI Summary](https://zoom-lfx.platform.linuxfoundation.org/meeting/99965231171-1782831600000/summaries?password=bf6b889f-a928-47f3-92d9-d631c7bb0cf9)
+
+## 👥 Attendees
+_(see Google Doc for attendees list)_
+
+## 📝 Quick Recap
+
+The meeting focused on developing a capabilities matrix for testing and evaluating different batch schedulers, which was previously discussed under the term "benchmarking." Marlow and Philip led the discussion about creating a framework that would allow users to test various scheduling capabilities across different workloads, starting with Kueue scheduler in Kubernetes. The group reviewed a document and Trello board containing different sections including infrastructure foundations, scheduler evaluation, and workload scenarios. Key participants including Kevin, Amy, and Abhishek provided feedback on how to structure the testing framework, with Kevin suggesting they focus first on defining how to use scheduling capabilities in Kueue and what workloads to support. The team agreed to start with workload scenarios and capabilities documentation, with plans to create a GitHub repository for actual implementation. James volunteered to reach out to Ricardo about potentially helping with the project, and Mesut offered to present about Project HAMI, a GPU slicing technology, at a future meeting.
+
+## ➡️ Next Steps
+
+- Filip: Share the Trello board link in the chat and add new members who request access.
+- Sabrina: Ensure the document link is publicly accessible and share it in the chat.
+- Marlow: Share the Alibaba data and the current working document in the channel.
+- Filip: Create tasks in the Trello board for the capability matrix sections and workload scenarios.
+- Kevin and Amy: Work on defining the workload scenarios and configurations for Kueue, including backfill and multi-cluster.
+- James: Speak with Ricardo Roca about involving the user group in filling out the capability matrix.
+- Filip: Add a "Future Directions" section to the document, including topics like auto-scaling and heterogeneous nodes.
+- Kevin: Prepare and give an overview of Workload Aware Scheduling (WASM) in about 4 weeks.
+- Mesut: Prepare to give an overview of the HAMI project at a future meeting.
+- All: Review the document and Trello board, and assign themselves to relevant tasks.
+
+## 📋 Summary
+
+### Meeting Attendance and Introductions
+
+The meeting began with participants joining and waiting for Alex and others to arrive. Marlow mentioned waiting for Alex to introduce new attendees and noted that Kevin would join later. The group discussed the presence of new faces and confirmed that no specific attendees were still awaited beyond Kevin and a potential CAI representative.
+
+### Capabilities Matrix Project Discussion
+
+The meeting focused on discussing a capabilities matrix project, which was previously referred to as benchmarking. Alex and Marlow noted that this conversation had been ongoing for about two years. The team also discussed logistics updates, including ongoing work on a white paper and meeting notes in GitHub. New participants were introduced, including Amy, Claudia, Killian, and others, who shared their backgrounds and current work in the batch and Kubernetes space. The conversation ended with plans to discuss testing-related topics, which were to be presented by Marlow or Philip.
+
+### Scheduler Performance Evaluation Planning
+
+The working group discussed adding scheduling capabilities and testing regimes for various schedulers to evaluate their performance across different algorithms and multi-node configurations. Marlow shared that Alibaba had developed a similar solution that could be leveraged, and Filip agreed to share the Trello board link for those interested in joining the project. The group noted they had recently pivoted in their approach and needed to update documentation, with Amy expressing interest in contributing to writing efforts once she better understood how to align her work with the project's goals.
+
+### Capabilities Suite for Scheduler Testing
+
+Marlow led a discussion about creating a capabilities suite to test different schedulers, particularly targeting Kueue for the first round implementation. The team discussed the need to organize and share documentation, with Filip offering to share screens and Sabrina confirming document access. Marlow explained that the suite would allow testing of various scheduling challenges including gang scheduling, resource fairness, queue management, preemption, and priority scheduling, with the goal of running the same workloads across different schedulers for comparative performance analysis.
+
+### Batch Scheduler Performance Testing
+
+The group discussed creating impartial performance tests for different batch schedulers to help users make informed decisions. Alex explained that while feature comparisons exist, there's a lack of objective performance data on factors like speed and scale capabilities. The team explored potential testing locations, including previous discussions with AWS representatives and other options, though they acknowledged the need to decide on specific testing environments. Abhishek suggested potentially using CNCF resources for CPU-based testing, though the group noted the importance of considering hardware limitations when evaluating different scheduling solutions.
+
+### Cloud-Native AI Scheduling Framework
+
+The team discussed developing a framework for benchmarking and evaluating scheduling capabilities for cloud-native AI workloads, with a focus on starting with Kueue scheduler implementation rather than trying to cover all schedulers simultaneously. They agreed to prioritize workload scenarios and capabilities documentation, with Kevin and Amy taking the lead on defining how to implement these capabilities in Kueue, while Marlow and Amy will handle the scheduler selection section. The group decided to keep the initial scope focused on a simpler pattern without auto-scaling for now, aiming to have something ready by KubeCon, and Mesut will present on Project HAMI (GPU slicing) at the next meeting.