AWS Bedrock Inference Profile Management Tool

Introduction

This tool provides a comprehensive solution for managing AWS Bedrock Application Inference Profiles. It allows you to create, list, and manage application inference profiles for Foundation Models or Cross-region Inference Profiles. The tool includes tagging capabilities and provides an interactive command line interface for easy management.

Features

• Create Application Inference Profiles with:
  • Foundation Model support
  • Cross-region Inference Profile support
• Batch create Application Inference Profiles from a yaml file
• Batch tag existing Application Inference Profiles from a yaml file
• Created profiles are exported to CSV automatically
• Tag management for inference profiles
• List existing Application inference profiles
• Delete existing profiles
• Interactive command-line interface
• Only support On-Demand Models, do not support Provisioned Models.

Prerequisites

• Python 3.9 and above
• AWS Account with appropriate permissions
• AWS credentials configured (either through AWS CLI profiles or access keys)

Workflow

┌─────────────────────────────────────────────────────────────────┐
│                    Installation & Setup                         │
│  1. Install dependencies                                        │
│  2. Configure AWS credentials                                   │
└────────────────────────┬────────────────────────────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────────────┐
│              Choose Your Approach                               │
└─────────────┬───────────────────────────────────┬───────────────┘
              │                                   │
              ▼                                   ▼
┌─────────────────────────────┐   ┌─────────────────────────────┐
│   Interactive Mode          │   │   Batch Mode                │
│   (Single Profile)          │   │   (Multiple Profiles)       │
└──────────┬──────────────────┘   └──────────┬──────────────────┘
           │                                  │
           │                                  ▼
           │                      ┌─────────────────────────────┐
           │                      │ Generate YAML Template      │
           │                      │ bedrock_batch_create_       │
           │                      │ profiles.py                 │
           │                      └──────────┬──────────────────┘
           │                                 │
           │                                 ▼
           │                      ┌─────────────────────────────┐
           │                      │ Review & Modify YAML        │
           │                      │ - Update profile names      │
           │                      │ - Adjust tags               │
           │                      │ - Add existing profiles     │
           │                      └──────────┬──────────────────┘
           │                                 │
           │                                 ▼
           │                      ┌─────────────────────────────┐
           │                      │ Batch Create Profiles       │
           │                      │ -f ./bedrock-profiles.yaml  │
           │                      └──────────┬──────────────────┘
           │                                 │
           └─────────────┬───────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────────────┐
│              Optional: Tag Existing Profiles                    │
│  -t ./bedrock-profiles.yaml                                     │
└────────────────────────┬────────────────────────────────────────┘
                         │
                         ▼
┌─────────────────────────────────────────────────────────────────┐
│              Manage Profiles                                    │
│  - List profiles (-l)                                           │
│  - Delete profiles                                              │
│  - Export to CSV (automatic)                                    │
└─────────────────────────────────────────────────────────────────┘

Installation

1. Clone the repository:

git clone <repository-url>
cd sample-bedrock-inference-profile-mgmt-tool

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows use: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Required AWS Permissions

When using the Application Inference Profile, ensure you have the following IAM permissions (replace <region> and <account_id> with your values):

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": ["bedrock:InvokeModel*"],
      "Resource": [
        "arn:aws:bedrock:<region>:<account_id>:inference-profile/*",
        "arn:aws:bedrock:<region>::foundation-model/*"
      ]
    }
  ]
}

Usage

Creating a New Inference Profile

Run the tool in interactive creation mode:

python bedrock_inference_profile_management_tool.py

The tool will guide you through:

1. AWS credential configuration (profile selection or manual input)
2. Tag configuration
3. Profile creation with options for:
   • Foundation Models
   • Cross-region Inference Profiles

Creating Batch Inference Profile YAML Template File

Generate a YAML template file with available models and inference profiles:

python3 bedrock_batch_create_profiles.py

The tool will:

1. Prompt you to select AWS profile and region
2. Ask for a filter keyword (e.g., claude, titan, llama, mistral)
3. List matching Foundation Models and System-Defined Inference Profiles
4. Generate a timestamped YAML file (e.g., claude-bedrock-profiles-20260115011234567.yaml)

Important: Before using the generated YAML file, you must:

• Review and modify profile names (especially for inference profiles that have "CHANGE ME-" prefix)
• Adjust tags as needed
• Optionally add existing profiles to tag under existing-profiles-to-tag section

Batch Creating Inference Profiles from a YAML File

Run the tool in batch creation mode:

python3 bedrock_inference_profile_management_tool.py -f ./bedrock-profiles.yaml

Batch Tagging Existing Inference Profiles from a yaml file

Run the tool in batch tagging mode:

python3 bedrock_inference_profile_management_tool.py -t ./bedrock-profiles.yaml

Listing and Managing Existing Profiles

To list and manage existing profiles:

python bedrock_inference_profile_management_tool.py -l

This command will:

1. Display all existing Application inference profiles
2. Allow you to delete selected profiles
3. Show profile details including:
   • Profile Name
   • Region
   • Model ID
   • Status
   • ARN
   • Associated Tags

CSV Export

The tool automatically exports profile information to CSV files with timestamps for record-keeping. The CSV includes:

• Profile Name
• Profile ARN
• Associated Tags

Error Handling

• The tool includes comprehensive error handling for common scenarios
• Provides clear error messages and retry options
• Validates inputs to prevent invalid operations

Verification

Check Tags on Inference Profile

Verify that tags have been applied to your inference profile:

aws bedrock list-tags-for-resource \
  --resource-arn arn:aws:bedrock:us-west-2:123456789:application-inference-profile/aaabbbccc \
  --region us-west-2

Example output:

{
    "tags": [
        {
            "key": "map-migrated",
            "value": "mig123Bedrock"
        },
        {
            "key": "Environment",
            "value": "Development"
        }
    ]
}

Check Invocation Success

Monitor invocations to verify the profile is being used successfully:

aws cloudwatch get-metric-statistics \
  --namespace AWS/Bedrock \
  --metric-name Invocations \
  --dimensions Name=ModelId,Value=aaabbbccc \
  --start-time 2026-01-14T02:25:00Z \
  --end-time 2026-01-14T02:35:00Z \
  --period 300 \
  --statistics Sum \
  --region us-west-2

Example output:

{
    "Label": "Invocations",
    "Datapoints": [
        {
            "Timestamp": "2026-01-14T02:30:00+00:00",
            "Sum": 1.0,
            "Unit": "Count"
        }
    ]
}

Contributing

Contributions are welcome! Please feel free to submit pull requests or create issues for bugs and feature requests.