I design evaluation-first AI systems for research, evidence-grounded analysis, and AI-native media production.
Current focus
Agentic Engineering (AE+X)
I treat AI systems as workflows that must be measured, audited, and improved — not just demoed. My core question is how to make multi-agent systems deliver reliable outcomes in real research and production settings.
- LLM-assisted qualitative coding, calibration, and failure analysis.
- Evidence-grounded generation and traceability for high-stakes AI use.
- Code-first media workflows spanning scripting, visuals, packaging, and publishing.
What I build
- Evaluation harnesses, datasets, and QA loops for human/LLM hybrid workflows.
- RAG and tool-using agent systems designed for reproducibility rather than demos.
- Teaching and studio formats that connect prototyping, documentation, and responsible AI practice.
Selected work
LLM reliability
Evaluation for complex qualitative coding
I build measurement protocols for LLM-assisted coding tasks, with emphasis on confidence, diversity, calibration, and targeted error correction.
Grounded AI
Verifiable evidence chains in health guidance
My recent work tests whether citations in patient-facing LLM outputs are actually traceable, auditable, and tied to the evidence they claim to represent.
Communication research
Media, health, and social impact
Alongside AI-systems work, I publish on digital media, framing, and communication effects across family, health, and regional media contexts.
Start here
Research
Project pipeline, preprints, and work in progress across evaluation, grounding, and communication research.
Browse research →Media
AI-native media projects, production workflows, and experiments in script-to-publish pipelines.
See media work →Digest
My running stream of model updates, agent tooling, product launches, and workflow-relevant technical signals.
Read the digest →Background
I am Zhilong George Zhao (赵志龙), a communication and digital media researcher working at the intersection of AI-native workflows, LLM evaluation, and multi-agent systems. My next 3–5 year agenda is Agentic Engineering (AE+X): designing, testing, and deploying multi-agent architectures that deliver reliable outcomes in real research and production workflows.
I am currently a Postdoctoral Researcher and Assistant Researcher at the School of Journalism and Communication, South China University of Technology (SCUT), and I am preparing faculty applications in Communication & Digital Media.
