Jan 2024 – Present · 2 yrs 5 mos

• 1. Worked on RAG-based chatbot solution to help the call center agents searching for procedure books in an efficient way while handling unfamiliar call scenarios, which results in low average call handling time, low call transfers and high customer satisfaction.
• Led the RAG pipeline model development process for a Generative AI powered Microsoft Teams based chatbot and collaborated with the technology partners to successful deployment resulting in 25% average call handling time (AHT).
• Implemented a RAG pipeline from scratch optimizing for context understanding and factual consistency.
• Collaborated with technology partners and implemented MLOps practices for RAG pipelines, including automated model retraining, versioning of contents and embeddings, and real-time performance monitoring.
• 2. AI chatbot utilizing a RAG architecture and AI agents to provide financial analysts and the executives with efficient and accurate search capabilities for deposit transaction summaries, peer financial documentations, improving data accessibility for strategic decision-making:
• Implemented GCP services like Document AI, function calling, GCP embedding models, Gemini model at various stages in the pipeline.
• Used Google Cloud Platform function calling capabilities and gemini-1.5-pro to convert natural language to SQL (text2SQL) queries to retrieve data from BigQuery tables

Aug 2021 – Jan 2024 · 2 yrs 5 mos

• As a part of AI ML COE team:
• 1. Developed an LLM based classification model to predict proper industry and sector classification code which results in Improved data quality and better risk assessment
• 2. Validated several other internal and vendor models for error root cause analysis, explainability analysis and sensitivity analysis.