
AI Engineer with 5+ years of experience building production-grade GenAI and AIOps systems, including LLM-based RAG pipelines, intelligent agents, and autonomous decision workflows. Strong expertise in end-to-end AI system design, optimization, and enterprise deployment.
Intelligent Customer Support Chatbot: Developed a Streamlit based chatbot using LangChain and large language models to simulate real customer interactions, improving engineer training efficiency by 30%. Created a RAG application with vector DB and GPT-4 for the LLM call. Integrated Sentence Transformers for real-time similarity analysis and dynamic conversation management. This enabled the creation of realistic training scenarios, helping new engineers improve their response accuracy and customer handling skills.,
Generative AI-Driven Customer Insights Platform: Developed a feedback analysis tool using FastAPI, React, and LangChain for prompt engineering, integrating Large Language Model (LLMs) to generate actionable insights from customer feedback, reducing manual analysis time by 40%. Designed a dynamic frontend to display AI-driven recommendations and issue analysis, empowering leaders with real-time insights, improving customer satisfaction by 20%, and deploying the solution to the cloud for scalability and seamless access.,
High Propensity Account Identification for POS Attachment: Performed data preprocessing and feature engineering to identify high propensity accounts for Point of Sale (POS) attachment, focusing on accounts with strong conversion potential but no current POS. Applied machine learning models to predict the likelihood of POS attachment, optimizing targeting strategies and driving a significant increase in POS attachment rates.
Identification of Online Auction Bidding Robots Using Machine Learning: International Conference on Evolutionary Computing and Mobile Sustainable Networks 2020. Selected as Lecture Notes on Data Engineering and Communications Technologies and as a chapter in book Evolutionary Computing and Mobile Sustainable Network.