Maniphest T433975

[draft] Q1 FY2026-27 LLM Platform Milestone 2
Open, Needs TriagePublic
Actions

Assigned To

None

Authored By

	isarantopoulos
	Aug 4 2026, 1:32 PM

Project Tags

Referenced Files

None

Subscribers

Related Objects
Search...

		Status	Subtype	Assigned	Task
		Open		None	T433975 [draft] Q1 FY2026-27 LLM Platform Milestone 2
		Open		kevinbazira	T433569 qwen36: Support thinking mode on the OpenAI-compatible endpoints
		Resolved		BWojtowicz-WMF	T431089 Set ephemeral-storage requests/limits on ML predictor containers to prevent node disk eviction
		Open		DPogorzelski-WMF	T433973 Update kserve helm chart to include LLMInferenceService
		Resolved		kevinbazira	T433977 Support OpenAI-compatible tool calling in the qwen36 model-server
		Open		BWojtowicz-WMF	T424923 Support GPU Partition specification in deployments
		Open		None	T436928 ml-serve-eqiad: Repartition two MI300X nodes from 24GB to 96GB partitions
		Open		kevinbazira	T434059 Enable Structured Outputs for LLMs
		Open		None	T429237 monitoring: Grafana dashboard for LLM serving on MI300X
		Open		None	T429236 monitoring: View GPU usage per LLM deployment/model

Event Timeline

isarantopoulos created this task.Aug 4 2026, 1:32 PM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptAug 4 2026, 1:32 PM

isarantopoulos added subtasks: T433569: qwen36: Support thinking mode on the OpenAI-compatible endpoints, T431089: Set ephemeral-storage requests/limits on ML predictor containers to prevent node disk eviction, T433973: Update kserve helm chart to include LLMInferenceService.Aug 4 2026, 1:35 PM

isarantopoulos added a subtask: T433977: Support OpenAI-compatible tool calling in the qwen36 model-server.Aug 4 2026, 1:37 PM

isarantopoulos added a subtask: T424923: Support GPU Partition specification in deployments.Aug 4 2026, 1:39 PM

isarantopoulos added a subtask: T434059: Enable Structured Outputs for LLMs.Aug 5 2026, 10:41 AM

kevinbazira closed subtask T433977: Support OpenAI-compatible tool calling in the qwen36 model-server as Resolved.Aug 7 2026, 8:14 AM

BWojtowicz-WMF closed subtask T431089: Set ephemeral-storage requests/limits on ML predictor containers to prevent node disk eviction as Resolved.Aug 11 2026, 12:51 PM

isarantopoulos added a subtask: T429237: monitoring: Grafana dashboard for LLM serving on MI300X.Fri, Sep 4, 5:13 AM

isarantopoulos added a subtask: T429236: monitoring: View GPU usage per LLM deployment/model.

isarantopoulos mentioned this in T431846: Q1 FY2026-27 Milestone 1: Deliver a V0 LLM platform for Wikimania.