Pushpendra Saini

Project: Generative AI-based Chatbot for Health Insurance Queries

• Developed a generative AI chatbot for health insurance client to answer policy-related questions. Extracted and processed policy documents, created embeddings using LangChain framework, and stored them in ChromaDB. Tested five open-source large language models for accurate summarization.
• Evaluated five LLM models, with Llama2 achieving the highest performance.
• Achieved 95% accuracy for FAQ-based questions and 80% accuracy for straightforward language queries.

Insights-Driven Chatbot for Business Queries

• Developed and enhanced a chatbot solution to respond to business queries with data visualizations (tables, charts, insights).
• Implemented data size reduction strategies prior to pipeline integration, improving efficiency.
• Created a benchmarking solution and monitored performance for LLM calls, ensuring optimal operation.
• Designed and implemented a dashboard for tracking performance metrics.
• Served as a Subject Matter Expert (SME) for the deployment of the chatbot solution across various projects.
• Improved code quality and functionality by implementing enhancements and resolving issues encountered by project owners.

• Captured approximately 5,000 high-quality images across various genres for digital projects and served as editor.
• Managed social media channels to enhance visibility and engagement.
• Prepared and presented weekly MIS reports to demonstrate growth in image contributions and sales data, utilizing MS Excel to create visual data presentations.