Streamlining Public Comment Analysis: USDA

Built a React-based Public Comment Analysis portal for USDA teams to submit Regulations.gov document IDs, track multi-stage processing in real time, and open a dedicated insights view to review clustered sentiment and themes.   

USDA receives thousands of public submissions on Regulations.gov. Manual categorizing and summarizing is slow, labor-intensive, and can miss key themes at scale. They needed automated topic modeling, sentiment analysis, AI-generated content detection, and accurate clustering/summarization.   

USDA Public Comment Analysis uses a serverless pipeline where the React (Amplify-hosted) UI submits a Regulations.gov document ID via API Gateway, triggers Lambda/Step Functions to ingest comments/attachments from S3, run clustering/sentiment/LLM insights, and store outputs for retrieval.   

On the frontend, I wired real-time status updates over WebSockets for live progress and a “Show Insights” view that fetches results via GET APIs for structured review.  I also implemented shared global state (documents, status, connection health) using React Context so the UI stays consistent across components as jobs move through the pipeline. To keep review fast for analysts, I structured the UI so they can jump from a completed run into a focused insights surface without losing their place in the processing queue. I treated the frontend like an operator console for a long-running pipeline, not a static dashboard. The UI had to make “async AI processing” feel legible, predictable, and easy to review.

Frontend: React (hooks), React Context, AWS Amplify hosting/deploy, REST + WebSocket integration.
Backend: API Gateway, Lambda, Step Functions, S3, DynamoDB, SageMaker clustering, Bedrock for insight generation, CDK

What I built

• A React UI for submitting document IDs and monitoring processing status, hosted on AWS Amplify with CI/CD from GitHub 

• Real-time progress UX powered by WebSocket messages, parsed client-side and reflected via React hooks/state updates   

• “Show Insights” flow that fetches analysis via GET and renders it in an overlay InsightPanel, including computed overall sentiment from cluster-level sentiment 

• Shared global state using React Context for document status, WebSocket connection status, and document lists 

Industry Impact

This AI-driven initiative is set to revolutionize public engagement and IT support. The Public Comment Analysis Tool will minimize the manual workload involved in reviewing public feedback, enabling faster insights and more data-driven policymaking. Through automated sentiment analysis and issue identification, the USDA can efficiently handle large volumes of comments, pinpoint key recommendations, and improve transparency in regulatory decisions.

For government agencies and businesses, these AI-powered solutions create a scalable approach to data analysis and IT automation, improving efficiency, engagement, and decision-making.

Potential for Wider Application 

Beyond the USDA, these AI-driven solutions can benefit various industries. Government regulatory agencies can use automated public comment analysis to accelerate policy decisions, while legal and compliance teams can categorize and analyze large-scale regulatory feedback. Corporate IT support can reduce workload with AI-powered self-service chatbots, and universities and research institutions can enhance public engagement and streamline feedback processing. By leveraging AI for automation and data-driven insights, these solutions improve efficiency, lower costs, and enhance both public and internal service experiences.