Streamlining Comment Insights

Leveraging Google Cloud Platform (GCP) services and large language models (LLMs), we provide clear summaries for each rating-based comment, analyze the sentiment of simpler comments, and deliver actionable insights based on the aggregated feedback.

Scope

Encompasses designing and implementing a solution to automate the analysis of user feedback from JSON files. This involves summarizing rating-based comments using advanced large language models (LLMs) to provide clear and concise insights, performing sentiment analysis on simple comments to deliver an overall summary, and developing a secure HTTPS REST API to process and present the data in a user-friendly JSON format. 

Our Contribution

• GCP Integration: Integrated GCP’s Text-Bison and Gemini Pro APIs with Langchain to handle summarization and sentiment analysis efficiently.
• Pipeline Development: Built a robust pipeline to manage interactions between text-bison and Gemini APIs for comprehensive data processing.
• API Endpoint Development: Developed and deployed a REST API using FastAPI and Uvicorn, ensuring high performance and scalability with Docker and GKE.
• Deployment: Dockerized the REST API, pushed it to Artifact Registry, and deployed it to a GKE cluster with a secure ingress endpoint for easy access.

Technical Solutions: Overcame GKE cluster connection issues by using a GCP bastion host and kubectl commands for secure and reliable access.