Skip to content

Instantly share code, notes, and snippets.

View threcc's full-sized avatar
🚀

Thomas Recchiuto threcc

🚀
View GitHub Profile
@threcc
threcc / create_modelmesh_workload.sh
Created March 12, 2026 14:34
Create a ModelMesh InferenceService with auth on RHOAI (S3 storage, OVMS, MNIST)
#!/bin/bash
#
# Creates a ModelMesh InferenceService with auth in the given namespace.
# Self-contained — no external manifest files needed.
# Uses S3 storage with a MNIST ONNX model.
# Model: MNIST-8 (handwritten digit classification, ONNX format)
# - Storage: S3 bucket (requires AWS credentials via env vars)
# - Runtime: OpenVINO Model Server (OVMS) in multi-model mode
#
# Prerequisites:
@threcc
threcc / create_serverless_workload.sh
Last active March 12, 2026 14:34
Create a serverless InferenceService with auth on RHOAI (OCI/ModelCar, OVMS, MNIST)
#!/bin/bash
#
# Creates a serverless InferenceService with auth in the given namespace.
# Self-contained — no external manifest files needed.
# Uses OCI (ModelCar) storage with a public MNIST ONNX model.
# Model: MNIST-8.1 (handwritten digit classification, ONNX format)
# - Image: quay.io/mwaykole/test (public, no auth required)
# - Input: 1x1x28x28 FP32 tensor (28x28 grayscale image)
# - Output: 1x10 FP32 tensor (digit probabilities 0-9)
# - Runtime: OpenVINO Model Server (OVMS)