Add example: deploy Faster-Whisper as a custom model on a SageMaker Triton MME

deploy_and_monitor/sm-triton_mme_faster_whisper/README.md

@@ -0,0 +1,45 @@
+# Deploying Pre-trained Faster-Whisper-Large-v3 Model on SageMaker with Multi-Model Endpoint and Triton Serve
+
+This guide provides an overview of deploying a pre-trained [Faster-Whisper-Large-v3 model](https://huggingface.co/Systran/faster-whisper-large-v3) from Hugging Face on Amazon SageMaker. The deployment utilizes a Multi-Model Endpoint, allowing efficient serving of multiple models on a single SageMaker instance.
+
+## Overview
+
+1. **Model Selection:** Choose the Faster-Whisper-Large-v3 model from Hugging Face's model hub.
+2. **SageMaker Deployment:** Deploy the selected model on Amazon SageMaker, a managed service for ML model training and deployment.
+3. **Multi-Model Endpoint:** Utilize a Multi-Model Endpoint on SageMaker for hosting and serving multiple models on a single instance.
+4. **Triton Serve Integration:** Integrate Triton Serve, an open-source model serving platform, for dynamic loading and unloading of models from the GPU.
+
+## Cost Optimization
+
+By using a Multi-Model Endpoint on SageMaker, achieve cost savings by maximizing resource utilization and cost efficiency.
+
+## Prerequisites
+
+Ensure [git-lfs](https://git-lfs.com/) is installed.
+
+### Install Dependencies
+
+```bash
+%pip install -qU pip awscli boto3 sagemaker
+```
+
+### Setting up SageMaker Execution Role
+
+1. Create a SageMaker execution role `AmazonSageMaker-ExecutionRole-mme-FasterWishper` with necessary policies.
+2. Assign the created role to your SageMaker instance/job.
+
+## Usage
+
+1. Clone the Faster-Whisper-Large-v3 model from Hugging Face.
+2. Arrange the model as Triton server expects it.
+3. Create the Triton server configuration file (`config.pbtxt`).
+4. Package the model as `*.tar.gz` for uploading to S3.
+5. Build and push the Docker image to ECR.
+6. Creat SageMaker Multi Model EndPoint base on Triton server
+7. Test your endpoint
+
+---
+
+This summarizes the key steps and commands involved in deploying the Faster-Whisper-Large-v3 model on SageMaker with a Multi-Model Endpoint and Triton Serve. Adjustments can be made based on specific requirements and configurations.
+
+Link to the [example notebook](sagemaker-mme-triton-fasterwhisper.ipynb)