Skip to main content

Deploying NudgeBee AI on AWS SageMaker

Overview​

This guide walks you through deploying the NudgeBee AI model on AWS SageMaker.

Prerequisites​

  • AWS account with access to SageMaker, S3, IAM, and ECR
  • Trained NudgeBee model (.tar.gz)
  • Model uploaded to an S3 bucket
  • IAM role with required permissions

Step-by-Step Guide​

Step 1: Upload Model to S3​

  1. Log in to AWS Console
  2. Go to Amazon S3
  3. Create or open a bucket
  4. Upload nudgebee_model.tar.gz to the bucket

Step 2: Create SageMaker Model​

  1. Open Amazon SageMaker
  2. Go to Models > Create model
  3. Name the model (e.g., nudgebee-ai-model)
  4. Provide ECR container image URL
  5. Add S3 path to the model artifact
  6. Choose IAM role with permissions
  7. Click Create model

Step 3: Deploy Endpoint​

  1. Go to Inference > Endpoint Configurations
  2. Create new configuration and add the model
  3. Choose instance type (e.g., ml.m5.large)
  4. Create endpoint and wait for it to become InService

RAG and LLM Server Configuration​

RAG Server (SageMaker)​

EMBEDDINGS_PROVIDER=sagemaker
EMBEDDINGS_PROVIDER_REGION=<AWS_SageMaker_Region>
EMBEDDINGS_PROVIDER_API_ENDPOINT=<SageMaker_Endpoint_URL>
EMBEDDINGS_MODEL_NAME=<Model_Name>

LLM Server (SageMaker)​

LLM_PROVIDER=sagemaker
LLM_PROVIDER_API_ENDPOINT=<SageMaker_Endpoint_URL>
LLM_PROVIDER_REGION=<AWS_SageMaker_Region>

Testing​

  1. Use SageMaker Console > Endpoints > Test
  2. Provide a JSON request and validate response