Lead LLM Training & Summarization Architect (CoTutor AI)
As the architect and builder of CoTutor AI at Wiingy, spearheaded the implementation of multi-step LLM workflows including live class transcription, summarization, and content generation. Oversaw AI tasks that convert tutor sessions into summaries, podcasts, key concepts, and flashcards by leveraging large language models (LLMs). Managed end-to-end process for preparing, labeling, and structuring educational session data to train and fine-tune these generative AI models. • Built and maintained text-based pipelines for summarizing educational session transcripts • Managed prompt-response workflows and RLHF data for tuning LLMs • Oversaw the extraction and curation of key concepts and flashcards from academic content • Used proprietary internal tools and LangChain/LangGraph for custom workflows