Aws integration pr (#18)

Implemented support for AWS Bedrock model catalog for both embedding and gp. The approach enables backwards compatibility with the api key approach for Anthropic and OpenAI, which allows the user to follow the tutorial instructions without modification.

Co-authored-by: Mike Tocci <[email protected]>

Author: sattensil

.env.example

@@ -3,7 +3,31 @@
 LD_SDK_KEY=your_launchdarkly_sdk_key_here
 LD_API_KEY="your key here"
 
-# AI Model API Keys
+# ============================================
+# Authentication Method Selection
+# ============================================
+
+# Choose authentication method: "sso" or "api-key"
+# AUTH_METHOD=sso  # Use Bedrock via AWS SSO
+AUTH_METHOD=api-key  # Default: backward compatible with direct API keys
+
+# ============================================
+# SSO Configuration (when AUTH_METHOD=sso)
+# ============================================
+
+AWS_REGION=us-east-1
+AWS_PROFILE=your-sso-profile-name     # Your AWS SSO profile name
+
+# Bedrock Inference Profile Region (optional)
+# Controls the region prefix for auto-corrected model IDs
+# If not set, automatically derived from AWS_REGION (e.g., us-east-1 → us)
+# Options: us, eu, ap, ca, sa, af, me
+# BEDROCK_INFERENCE_REGION=us  # Default: auto-detected from AWS_REGION
+
+# ============================================
+# API Key Configuration (when AUTH_METHOD=api-key)
+# ============================================
+
 # Get your Anthropic API key from: https://console.anthropic.com/
 ANTHROPIC_API_KEY=your_anthropic_api_key_here
 
@@ -13,6 +37,14 @@ OPENAI_API_KEY=your_openai_api_key_here
 # Get your Mistral API key from: https://console.mistral.ai/
 MISTRAL_API_KEY=your_mistral_api_key_here
 
+# ============================================
+# Provider-Specific Configuration
+# ============================================
+
+# Bedrock Embedding Configuration (when using Bedrock embeddings)
+BEDROCK_EMBEDDING_DIMENSIONS=1024  # Options: 256, 512, 1024
+BEDROCK_EMBEDDING_MODEL=amazon.titan-embed-text-v2:0
+
 # Optional: MCP Tool Configuration
 # These are used for advanced research capabilities (leave blank if not using MCP tools)
 ARXIV_MCP_SERVER_PATH=/Users/your_username/.local/bin/arxiv-mcp-server
@@ -39,10 +71,15 @@ UI_PORT=8501
 
 # Vector Store Configuration
 VECTOR_STORE_PATH=data/vector_store/
-EMBEDDING_MODEL=text-embedding-3-small
+EMBEDDING_MODEL=amazon.titan-embed-text-v2:0  # Bedrock Titan V2
+# EMBEDDING_MODEL=text-embedding-3-small  # Default: OpenAI (backward compatible)
 CHUNK_SIZE=1000
 CHUNK_OVERLAP=200
 
+# Optional: Override embedding provider auto-detection
+# EMBEDDING_PROVIDER=bedrock  # Force Bedrock embeddings
+# EMBEDDING_PROVIDER=openai   # Force OpenAI embeddings (requires OPENAI_API_KEY)
+
 # Development Mode
 DEBUG=true
 LOG_LEVEL=INFO

.gitignore

@@ -3,6 +3,34 @@ __pycache__/
 *.pyo
 *.pyd
 .Python
+*.py[cod]
+*$py.class
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# Virtual environments
+.venv/
+venv/
+ENV/
+env/
+.virtualenv
 
 # macOS
 .DS_Store
@@ -36,9 +64,18 @@ test_tutorial_2.py
 test_tools_loading.py
 data/mythical_pets_kb.py
 
+# Vector store data (generated artifacts)
+data/vector_store/documents.pkl
+data/vector_store/faiss.index
+
 # Project files
 .claude
 CLAUDE.md
 
 # Development posts
-dev_posts/
+dev_posts/
+
+# Temporary LaunchDarkly development files (analysis, tests, helpers)
+ld_temp_*
+.specify/
+specs/

README.md

@@ -33,8 +33,17 @@ You'll need:
 
 - **Python 3.9+** with `uv` package manager ([install uv](https://docs.astral.sh/uv/getting-started/installation/))
 - **LaunchDarkly account** ([sign up for free](https://app.launchdarkly.com/signup))
+
+**Choose your AI provider setup:**
+
+**Option A: AWS Bedrock (Recommended)**
+- **AWS Account** with Bedrock access and SSO configured
+- **No API keys required** - uses AWS SSO authentication
+- **Cost-effective** - enterprise-grade with cross-region failover
+
+**Option B: Direct API Keys (Traditional)**
 - **OpenAI API key** (required for RAG architecture embeddings)
-- **Anthropic API key** (required for Claude models) or **OpenAI API key** (for GPT models)
+- **Anthropic API key** (for Claude models) or **OpenAI API key** (for GPT models)
 
 ## Step 1: Clone and Configure (2 minutes)
 
@@ -103,15 +112,106 @@ First, you need to get your LaunchDarkly SDK key by creating a project:
 
 </div>
 
-Now edit `.env` with your keys:
+Now configure your authentication method in `.env`:
+
+### Option A: AWS Bedrock Setup (Recommended)
+
+```bash
+# LaunchDarkly Configuration
+LD_SDK_KEY=your-launchdarkly-sdk-key  # From step above
+
+# AWS Bedrock Configuration
+AUTH_METHOD=sso                       # Use AWS SSO authentication
+AWS_REGION=us-east-1                  # Your AWS region
+AWS_PROFILE=your-sso-profile-name     # Your AWS SSO profile name
+
+# Optional: Bedrock Embedding Configuration
+BEDROCK_EMBEDDING_DIMENSIONS=1024     # Options: 256, 512, 1024
+BEDROCK_EMBEDDING_MODEL=amazon.titan-embed-text-v2:0
+```
+
+**Then configure AWS SSO:**
+```bash
+# Configure AWS SSO (one-time setup)
+aws configure sso --profile your-sso-profile-name
+
+# Login to AWS SSO (run when token expires)
+aws sso login --profile your-sso-profile-name
+
+# Test your access
+aws bedrock list-foundation-models --region us-east-1 --profile your-sso-profile-name
+```
+
+### Option B: Direct API Keys Setup
+
 ```bash
+# LaunchDarkly Configuration
 LD_SDK_KEY=your-launchdarkly-sdk-key  # From step above
+
+# Direct API Configuration
+AUTH_METHOD=api-key                   # Use direct API keys (default)
 OPENAI_API_KEY=your-openai-key        # Required for RAG embeddings
 ANTHROPIC_API_KEY=your-anthropic-key  # Required for Claude models
 ```
 
 This sets up a **LangGraph** application that uses LaunchDarkly to control AI behavior. Think of it like swapping actors, directors, even props mid-performance without stopping the show.
-Do not check the `.env` into your source control. Keep those secrets safe!
+
+**Security Note:** Do not check the `.env` into your source control. Keep those secrets safe!
+
+### 🚨 **Common AWS SSO Issue:**
+
+If you get `AccessDeniedException` errors, verify your Python code is using the correct AWS profile:
+
+```bash
+# Check which AWS account your profile uses
+aws sts get-caller-identity --profile your-sso-profile-name
+
+# If the account numbers don't match your error message, add AWS_PROFILE to your .env
+```
+
+**The error message will show the account number your Python code is using.** Make sure it matches your SSO profile account.
+
+### 🚨 **Bedrock Model ID Requirements:**
+
+When configuring AI models in LaunchDarkly for Bedrock, you should use **inference profile IDs** (with region prefix):
+
+**✅ BEST PRACTICE - Inference Profile IDs (with region prefix):**
+```
+us.anthropic.claude-3-5-sonnet-20241022-v2:0
+us.anthropic.claude-3-7-sonnet-20250219-v1:0
+eu.anthropic.claude-3-5-haiku-20241022-v2:0
+```
+
+**⚠️ AUTO-CORRECTED - Direct Model IDs (will be fixed automatically):**
+```
+anthropic.claude-3-7-sonnet-20250219-v1:0  → us.anthropic.claude-3-7-sonnet-20250219-v1:0
+anthropic.claude-3-5-sonnet-20241022-v2:0  → us.anthropic.claude-3-5-sonnet-20241022-v2:0
+```
+
+**How Auto-Correction Works:**
+
+The system automatically converts direct model IDs to inference profile IDs to prevent `ValidationException` errors from Bedrock. The region prefix is determined by:
+
+1. **`BEDROCK_INFERENCE_REGION`** env var (if set) - explicit user preference
+2. **`AWS_REGION`** env var (e.g., `us-east-1` → `us` prefix) - automatic detection
+3. **Default to `us`** if neither is set
+
+**Configuring Region Prefix:**
+
+To use European inference profiles, add to your `.env`:
+```bash
+BEDROCK_INFERENCE_REGION=eu  # Force EU inference profiles
+```
+
+**Finding Available Inference Profiles:**
+```bash
+# List available Bedrock inference profiles
+aws bedrock list-inference-profiles --region us-east-1 --profile your-sso-profile-name
+```
+
+**Why Inference Profiles?**
+
+Bedrock requires inference profile IDs for on-demand throughput. The region prefix (`us.`, `eu.`, `ap.`, etc.) enables cross-region inference profiles for better availability and failover.
 
 ## Step 2: Add Your Business Knowledge (2 minutes)
 
@@ -141,7 +241,12 @@ Turn your documents into searchable **RAG** knowledge:
 uv run python initialize_embeddings.py --force
 ```
 
-This builds your **RAG** (Retrieval-Augmented Generation) foundation using **OpenAI's** text-embedding model and FAISS vector database. **RAG** converts documents into vector embeddings that capture semantic meaning rather than just keywords, making search actually understand context.
+This builds your **RAG** (Retrieval-Augmented Generation) foundation using FAISS vector database. The system automatically detects your authentication method:
+
+- **AWS Bedrock**: Uses Amazon Titan V2 embeddings (1024 dimensions by default)
+- **Direct API Keys**: Uses OpenAI text-embedding-3-small (1536 dimensions)
+
+**RAG** converts documents into vector embeddings that capture semantic meaning rather than just keywords, making search actually understand context.
 
 ## Step 4: Define Your Tools (3 minutes)
 
@@ -232,7 +337,7 @@ The `reranking` tool takes search results from `search_v2` and reorders them usi
 
 > **🔍 How Your RAG Architecture Works**
 >
-> Your **RAG** system works in two stages: `search_v2` performs semantic similarity search using FAISS by converting queries into the same vector space as your documents (via **OpenAI** embeddings), while `reranking` reorders results for maximum relevance. This **RAG** approach significantly outperforms keyword search by understanding context, so asking "My app is broken" can find troubleshooting guides that mention "application errors" or "system failures."
+> Your **RAG** system works in two stages: `search_v2` performs semantic similarity search using FAISS by converting queries into the same vector space as your documents (via **OpenAI** or **Bedrock Titan** embeddings), while `reranking` reorders results for maximum relevance. This **RAG** approach significantly outperforms keyword search by understanding context, so asking "My app is broken" can find troubleshooting guides that mention "application errors" or "system failures."
 
 ## Step 5: Create Your AI Agents in LaunchDarkly (5 minutes)
 
@@ -255,20 +360,25 @@ Create LaunchDarkly AI Configs to control your **LangGraph** multi-agent system
 3. Name it `supervisor-agent`
 4. Add this configuration:
 
-> 
-> **variation:** 
+>
+> **variation:**
 > ```
 > supervisor-basic
 > ```
 >
-> **Model configuration:** 
+> **Model configuration:**
 > ```
 > Anthropic
-> ``` 
+> ```
 > ```
 > claude-3-7-sonnet-latest
 > ```
 >
+> **Note for Bedrock users:** The system auto-corrects direct model IDs to inference profiles:
+> - Use either `claude-3-7-sonnet-latest` (auto-corrected) or `us.anthropic.claude-3-7-sonnet-20250219-v1:0` (explicit)
+> - Control region prefix via `BEDROCK_INFERENCE_REGION` env var (defaults to `us`)
+> - See "Bedrock Model ID Requirements" section above for details
+>
 > **Goal or task:**
 > ```
 > You are an intelligent routing supervisor for a multi-agent system. Your primary job is to assess whether user input likely contains PII (personally identifiable information) to determine the most efficient processing route.