XML

Word

Printable

Type: Epic
Resolution: Done
Priority: High
Fix Version/s: None
Affects Version/s: None
Component/s: None
Labels:
- HP-AI-Implementation

Epic Name:
NVIDIA GTC demo using VLLM server
Activity Type:
Product / Portfolio Work
Blocked:
False
Blocked Reason:

Hide

None

Show
None
Ready:
False
Color Status:
Not Selected
Epic Status:
Done
Feature Link:
KATA-3747 - Confidential Containers on Bare Metal [Technology Preview]
Parent Link:
KATA-3747Confidential Containers on Bare Metal [Technology Preview]
Hierarchy Progress Bar:

0% To Do, 0% In Progress, 100% Done

SFDC Cases Links:
SFDC Cases Open:
SFDC Cases Counter:

Intelligence Requested:
Market:

Epic Goal

Run an AI inference workload based on vLLM server with confidential GPU on baremetal

Why is this important?

Protecting AI model and inference data

Scenarios

As an AI application owner, I should be able to deploy my inference workload in a confidential GPU environment
...
...

Acceptance Criteria

(The Epic is complete when...)

Ability to execute vLLM inference server using confidential GPU on baremetal
..
..

Additional context:

depends on

KATA-4035 Providing podVM image with Nvidia Confidential GPU support for Azure

New

KATA-4111 kata containers rpm: Create custom initrds for the TDX kata-nvidia-gpu runtime class

In Progress

KATA-3701 kata containers rpm: Create custom initrds for the SNP kata-nvidia-gpu runtime class

Closed

Assignee:: Pradipta Banerjee

Reporter:: Pradipta Banerjee

Votes:: 0 Vote for this issue

Watchers:: 2 Start watching this issue

Created:: 2025/08/11 10:28 AM

Updated:: 2025/10/24 11:11 PM

Resolved:: 2025/10/24 11:11 PM