Build and Deploy an MCP-Powered AI Workflow Using FastAPI and AWS

We are looking for a backend engineer to build and deploy a small AI workflow using the Model Context Protocol (MCP). Workflow The application should allow a user to ask a question about business data stored in an existing Flask application. The MCP workflow will: 1. Receive a user request through a FastAPI endpoint. 2. Allow an LLM to call an MCP tool. 3. Retrieve the required data from an existing Flask API. 4. Return a clear, structured response to the user. 5. Log the request, tool execution, response status, and failures. Example request: “Show me the customers who require follow-up this week.” The MCP tool should securely call the Flask API, retrieve the relevant records, and provide the results to the LLM. Technical Scope * Build a FastAPI service that acts as the AI and MCP client layer. * Create one custom MCP server or MCP tool. * Integrate the MCP tool with an existing Flask REST API. * Connect the workflow to OpenAI, Anthropic, or another supported LLM. * Add authentication between FastAPI and Flask. * Add input validation, timeouts, retries, and error handling. * Dockerize the services. * Deploy the application to AWS. * Add basic logging and monitoring. Preferred AWS Deployment The solution may use: * AWS ECS Fargate or EC2 * Amazon ECR * Application Load Balancer * AWS Secrets Manager * Amazon CloudWatch * PostgreSQL or an existing business database Deliverables * FastAPI application * Custom MCP server or tool * Flask API integration * LLM tool-calling workflow * Dockerfile and Docker Compose configuration * AWS deployment configuration * Environment and secrets configuration * Logging and error handling * Unit and integration tests * README with local setup and deployment instructions * Short recorded or live demonstration Required Experience * Strong Python experience * FastAPI and Flask * REST API development * AWS deployment * Docker * LLM tool calling * MCP server or client development * Authentication and secure credential handling Acceptance Criteria The project will be considered complete when: * A user can submit a request through the FastAPI endpoint. * The LLM correctly selects and invokes the MCP tool. * The MCP tool retrieves data from the Flask API. * The final response accurately reflects the returned business data. * Authentication, failure handling, and logging work correctly. * The application can be deployed and tested successfully on AWS. Please include examples of similar FastAPI, Flask, AWS, MCP, or LLM integration work in your proposal.