Niket Patil

Prompt Engineering for LLM-Powered E-Learning Content Creation Automation. Developed an automated Research Notes Generation agent that retrieves and synthesizes learning-objective-aligned research notes from multi-source relevant references using LLMs. Developed an automated Slide Content Generation agent that converts research notes into structured slide chunks using LLMs. Developed an automated Assessment Generation agent that creates assessment questions from slide content using LLMs. Developed an automated Graphics Definition Generation agent using LLMs to convert slide content into scene-based graphics definitions. Developed an automated Image Description Generator agent that analyzes educational images using multimodal LLMs to produce metadata.

Machine Learning for Large Language Models (LLM). Created a voice based chatbot streamlit app which is able to take interview of users and provide feedback of the user's performance in the interview. Created a Conversational Financial Advisor Chatbot, utilizing OpenAI's language models and contextual memory to provide personalized loan guidance and support for bank loans.

Machine Learning for Natural Language Processing (NLP). Designed a Sentiment Analysis (classification) model to improve customer service and support experience using Logistic Regression model on 6 lakh sentences, with an accuracy of 88%. Designed a Voice Activity Detection (VAD) model for voice-controlled systems and virtual assistants, integrating it into a Gradio interface to process audio files and remove silent segments.