Shibu Mohapatra

Took ownership of end-to-end development and deployment of Rima, a personal care retail chatbot leveraging LLMs, prompt engineering, Logfire, real-time conversational streaming, personalization, and Algolia-based product search. Built the core AI backbone using MongoDB, AWS S3, Redis, Celery, VectorDB-backed FAQ support, authentication API, a golden-coupon series feature, and conducted Locust load testing to support high-traffic events. Led prompt engineering for Agastya, an Astrology-based Agentic AI at AstroSure.ai, contributing to engagement growth across 350K+ MAUs within a 1M+ user base and integrating a subscription model. Spearheaded QA teams on reliability testing and system validation.

Engineered an LLM-powered chatbot using LangChain and SQLAlchemy to enable natural language querying of internal databases, reducing data retrieval time for non-technical teams. Launched a real-time generative AI chatbot using PyTorch and WebRTC, integrating HeyGen for avatar-based conversational interaction.

Created an NLP sentiment analysis model for a retailer, achieving 94% accuracy. Used ML algorithms to derive a best-fit model and classify phishing activities. A Deep Learning model (CNN, DNN, LSTM) to benchmark performance on the MNIST dataset.

Collaborated with cross-functional teams to develop BI Finance Bots using Python and Automation Anywhere, saving 8+ manual hours per report through automated GW dataset generation. Integrated Oracle Fusion, Oanda foreign exchange, Vertex global tax, and Coupa P2P data pipelines to streamline bot execution and improve revenue reporting accuracy.