Shubhapriya Ghosh

Built and deployed Looker Studio dashboards tracking real-time marketing KPIs and ROI. Developed a RAG-based conversational AI system using LLMs and vector search to automate customer engagement. Led and deployed a scalable RAG-based AI system using LangChain, Pinecone, and FastAPI, integrated with LLMs and HuggingFace embeddings on AWS with Docker. Optimized retrieval using hybrid search and re-ranking.

Architected an end-to-end data pipeline combining classical NLP and LLM-based reasoning for unstructured data. Led development of an NLP-based parsing system to extract structured information. Developed a GPT-3.5-powered question-answering layer to enhance resume understanding.

Analyzed large-scale image and user interaction data to improve virtual fitting accuracy. Leveraged trade-offs across multiple solution approaches. Optimized system performance metrics and collaborated with the Virtual Fitting product team to integrate models into production.

Contributed to the development of a Sign Language Recognition system. Implemented data preprocessing and feature representation pipelines. Assisted in model training and evaluation, achieving 94% classification accuracy on a benchmark dataset.