AI Trainer – LLM Dialogue and Text Annotation
I developed AI-driven agents to generate, analyze, and summarize textual data in response to user input. This involved curating and evaluating text datasets to enhance generative AI chatbot performance for safety consulting and event reporting. I focused on improving natural language outputs with prompt engineering and human-in-the-loop feedback. • Annotated and evaluated dialogues for generative assistant safety consultants using Llama and Gemini models. • Implemented, tested, and improved prompt + response interactions in a Streamlit dashboard. • Integrated feedback loops with memory checkpointing for refining responses and scenario handling. • Managed text data preprocessing and cleansing with Pandas and custom Python scripts.