Integrated opentelemetry.trace into the /audit endpoint to provide deep visibility into the video compliance workflow. This update wraps the asynchronous compliance_graph.ainvoke call in a custom span titled Execute-Compliance-Graph, enabling Azure Application Insights to group all internal logs, errors, and metadata under a single transaction.

Updated the telemetry.py configuration to improve visibility within Azure Application Insights by explicitly connecting the api-server logger to the OpenTelemetry pipeline. This ensures that internal application logs are correctly captured and correlated with incoming requests. Additionally, I exported a global tracer to facilitate manual span creation, allowing for granular tracking of the LangGraph workflow execution. These changes bridge the gap between automatic HTTP instrumentation and internal logic, providing a cohesive end-to-end transaction view in the Azure Portal, complete with custom attributes for session and video metadata.

Synchronous execution has been modified into Asynchronous execution

Updating README.md file

Integrated Azure Monitor OpenTelemetry to establish production-grade observability for the TubeGuardAI pipeline, enabling automated tracking of FastAPI requests, Azure AI Search queries, and internal logging events. The implementation includes a robust configuration utility that retrieves connection strings from environment variables, handles initialization exceptions, and provides clear diagnostic logging to verify telemetry status. This setup ensures full-stack visibility into system performance and request latency by connecting the application directly to Azure Application Insights.

Implemented a production-grade FastAPI backend layer to expose the LangGraph compliance workflow as a scalable REST API, featuring Pydantic models for strict request/response validation, unique session tracking via UUIDs, and centralized logging. The integration includes a dedicated telemetry setup for Azure Application Insights and provides a structured interface to orchestrate the transition from raw video URLs to deterministic JSON compliance reports, complete with an automated health-check endpoint.

Minor change

Commiting the updated .gitignore file

Changes made in the code to use the correct deployment name, endpoint, api key and api version for AzureOpenAIEmbeddings

Changes made in the code to use the correct deployment name, endpoint, api key and api version

Modified the script to load the Identity credentials from .env file for DefaultAzureCredential to generate access tokens.

Modified the comments for better understanding of the script

fixed a minor error

This update introduces the main execution entry point for TubeGuardAI, serving as the central orchestration layer for the compliance audit workflow. The script initializes the environment, generates unique session tracking via UUIDs, and manages the end-to-end lifecycle of a video audit—from input payload preparation to the invocation of the LangGraph-based brain. It features integrated logging for execution tracking, robust error handling to capture workflow failures, and a CLI-based reporting interface that transforms the final state into a structured compliance report detailing audit status, categorized violations by severity, and AI-generated summaries.

This update implements a robust RAG (Retrieval-Augmented Generation) indexing pipeline that automates the ingestion of local PDF documents into Azure AI Search. The script features a structured workflow that begins with environment validation and the initialization of Azure OpenAI Embeddings for vectorization, followed by a document processing layer that utilizes PyPDFLoader and RecursiveCharacterTextSplitter to transform raw text into manageable, context-aware chunks. By tagging each segment with source metadata and leveraging the AzureSearch vector store for batch uploads, this service establishes a high-performance knowledge base designed for efficient semantic search and accurate citation retrieval.