Shefali Goyal

Worked on large language models (LLM) and prompt engineering to develop AI-driven solutions for automating template filling processes. Built an Agentic AI system using Amazon Bedrock for document processing automation, deploying 4 specialized agents for scope checking, document classification, elementary data extraction, and metadata generation. Built automated Outlook email workflows using MCP servers and LangGraph, enabling dynamic tool-orchestrated processing and reducing manual email handling time.

Developed and implemented techniques for detecting and mitigating bias in the Medical Expenditure Panel Survey (MEPS) dataset. Trained and optimized machine learning models, including Logistic Regression and Random Forest, to identify and correct biases, improving the accuracy and fairness of predictive outcomes in the dataset.

Utilized Natural Language Processing (NLP) methods, including tokenization, lemmatization, and stopword removal, to preprocess a large dataset of TPDDL tweets for sentiment analysis. Employed machine learning techniques, such as Logistic Regression and Support Vector Machines (SVM), to classify tweets into sentiment categories like positive, negative, and neutral.