● Utilized Flask and FastAPI to Develop LLM APIs utilizing chain-of-thought prompting to generate fine-grained results and integrated with vector databases for efficient document retrieval using various approximate nearest neighbors (ANN) algorithms, such as HNSW (Hierarchical Navigable Small World) to efficiently retrieve relevant information. ● Developed web-based applications as the frontends for LLM APIs, enabling users to interact with our models through a user-friendly interface, providing real-time access to fine-grained results generated by the LLM APIs. ● Designed and implemented machine learning models for top financial institutions using XGBoost and Random Forest algorithms, improving classification accuracy by up to 50%. ● Leveraged Pandas to wrangle large datasets for model ingestion by cleaning and transforming using techniques like data imputation, normalization, and feature engineering. ● Applied NLP techniques to enhance signal in transcripts for classification, utilizing tokenization, lemmatization, and named entity recognition (NER). ● Trained and tested various machine learning and deep learning models for classification, entity recognition, and content generation. ● Communicated detailed data and model performance insights using visualization libraries like Seaborn & Matplotlib