Architected and deployed a full-stack AI assistant for the company website using Google Vertex AI (Gemini 2.0 Flash) and a Retrieval-Augmented Generation (RAG) pipeline, enabling natural language querying and grounding responses in real company documents. Built a Python FastAPI RAG service using LangChain to chunk and embed 250+ document sections (xlsx, docx, pdf) into a ChromaDB vector store, powered by Vertex AI text-embedding-004 for semantic search and retrieval. Developed a Node.js Express API gateway with session management, rate limiting, and input validation serving as the middleware layer between the frontend chat UI and the Vertex AI backend. Designed a pixel-perfect dark-themed chat UI with features like suggestion chips, typing indicators, RAG source attribution panels, and a styled fallback message system with email and Book-a-Call CTAs. Enriched the chatbot knowledge base with 3PL and supply chain content covering warehousing strategies, fulfillment KPIs, shipping zones,

Data Science Analyst at Akira Consultancy

February 1, 2025 - November 1, 2025

Client: Kantar - Advertising & Consumer Analytics. Advertising Platform Analytics: Quantified key drivers of advertiser spend using dominance analysis; applied correlation-based feature reduction to address multicollinearity, leveraged Cholesky decomposition for stability, and used SVD for factor extraction, prioritizing spend drivers and delivering prioritized platform attributes through automated Excel dashboards. Ad & Video Analytics – Creative Driver Identification: Modeled audience reach and engagement using OLS and Lasso regression on creative elements (product display, problem framing, and other attributes), identified high-impact ad levers with Link AI to optimize creative effectiveness.

Data Science Analyst at Akira Consultancy

September 1, 2024 - February 1, 2025

Client: Trent. Web scraping on Dmart and price benchmarking across multiple retailers; regional price comparisons; built ARIMA forecasting model on one year of Dmart data achieving 88% accuracy; optimized scraping pipeline with multiprocessing to reduce runtimes and enable daily automated file generation; deployed scripts on VM with Windows Task Scheduler for daily execution; designed Tableau dashboards to visualize price comparisons, regional trends, forecasting results, and presented insights to clients.

Associate Software Engineer at BNP Paribas

August 1, 2022 - August 1, 2023

Chennai, India (Onsite). Ensured accuracy of economic P&L figures for Global Markets across US, Europe & Asia, collaborating with Front Office and BA for the transactions recorded. Leveraged SQL to streamline data extraction and analysis, reducing time by 20% and delivering timely insights. Optimized Perl scripts and stored procedures, increasing data processing efficiency by 40%. Engineered ETL workflows in Informatica PowerCenter, reducing data loading time by 25%. Applied advanced troubleshooting techniques to resolve complex issues raised by Business Analysts.