kserve
Go Mediumkserve/kserve
5,285 stars
1,423 forks
26 open issues
Active Apr 2026
Beginner-Friendly Issues 26
Issues tagged for new contributors
Add llama.cpp as a CPU-optimized ServingRuntime for LLM inference
#5334 · Apr 2, 2026
Huggingface models are unusable
#5326 · Apr 1, 2026
kind/bug bug/regression
[Bug] Hardcoded namespace in `kserve-llmisvc-crd` chart prevents flexible deployment
#5325 · Mar 31, 2026
good first issue
HTTPRoute timeouts field breaks GKE Gateway controller — need option to disable
#5311 · Mar 27, 2026
Replace kserve.storage with OpenDAL
#5306 · Mar 26, 2026
ClusterStorageContainer is completely ignored
#5299 · Mar 24, 2026
Expose inference protocol and model name in InferenceService status
#5297 · Mar 23, 2026
MCP Server for KServe — AI-Native Operational Interface
#5295 · Mar 23, 2026
Proposal: Add DRA (Dynamic Resource Allocation) Support for GPU Management
#5294 · Mar 21, 2026
Proposal: NVIDIA Dynamo Runtime Integration
#5293 · Mar 21, 2026
Proposal: SGLang ClusterServingRuntime Integration
#5292 · Mar 21, 2026
Support Open Responses spec for inference API
#5285 · Mar 19, 2026
Helm chart v0.17.0 is missing
#5272 · Mar 17, 2026
Guided KServe install mission in KubeStellar Console
#5270 · Mar 17, 2026
No `./install/v0.17.0/` directory present in the `v0.17.0` kserve release
#5255 · Mar 14, 2026
kind/bug
Support per-LLMInferenceService resource overrides for storage initializer container
#5225 · Mar 11, 2026
Does it support Sglang PD separate deployment?
#5224 · Mar 11, 2026
huggingface base image has aged out, contains vulnerabilities
#5204 · Mar 9, 2026