A retrieval-augmented, LLM-powered news agent that fetches, compares, and summarizes multi-source articles into publish-ready scripts for short-form video content, cutting manual production time. Built with INIT Build, where I co-led development of the agentic workflow and data/API layers.
Built a RAG, LLM-powered agent that fetches, compares, and summarizes multi-source news into publish-ready video scripts.
Designed and normalized a Postgres schema in Supabase for video metadata, scripts, and storage assets.
Orchestrated a multi-step agentic workflow in LangGraph for reliable execution under loosely defined requirements.