Agentic AI Auto-Loan Platform
Budget: $15 – $25 USD
I want a compact yet fully functional auto-loan processing proof-of-concept built with an agentic AI approach. The workflow must cover the entire cycle: a Python-based loan application form, an AI-driven credit-scoring system, automated loan-approval logic, and a clear risk-analysis report that borrowers and underwriters can both understand.
The intelligence layer should rely on AWS Bedrock for orchestration and Anthropic Claude Code for the reasoning steps, while Amazon SageMaker handles model fine-tuning and scalable inference. The agent(s) you design must independently perform credit scoring, trigger approval decisions, and generate the risk summary without manual intervention.
Deliverables
• Clean, well-documented Python code that spins up the form, invokes Bedrock and SageMaker, and returns decisions in real time.
• A minimal front-end or API endpoint demonstrating the end-to-end flow.
• Sample data plus instructions so I can replicate the scoring and approval process in my own AWS account.
• A short README explaining how each agent operates, how Bedrock and SageMaker are wired together, and how to extend the framework for production.
Acceptance will be based on the application’s ability to accept a sample loan request, compute a score, decide approval, and output the accompanying risk report automatically.
The intelligence layer should rely on AWS Bedrock for orchestration and Anthropic Claude Code for the reasoning steps, while Amazon SageMaker handles model fine-tuning and scalable inference. The agent(s) you design must independently perform credit scoring, trigger approval decisions, and generate the risk summary without manual intervention.
Deliverables
• Clean, well-documented Python code that spins up the form, invokes Bedrock and SageMaker, and returns decisions in real time.
• A minimal front-end or API endpoint demonstrating the end-to-end flow.
• Sample data plus instructions so I can replicate the scoring and approval process in my own AWS account.
• A short README explaining how each agent operates, how Bedrock and SageMaker are wired together, and how to extend the framework for production.
Acceptance will be based on the application’s ability to accept a sample loan request, compute a score, decide approval, and output the accompanying risk report automatically.