Exploring Phoenix for Enhanced Research in Humanities and Social Sciences

Hi everyone! I lead a research software engineering (RSE) team at the Australian National University. We build software to support research in the computational humanities and social sciences. I'm a historian of ideas and technology by trade but have spent a couple of decades working with engineering teams. We're using Phoenix to capture telemetry and user feedback for a RAG-based tool for historical research and are considering how it could be used for more data intensive social science use cases. I'm interested in how tools like Phoenix can improve reproducibility and transparency of LLM-enabled research workflows.