Shuhao Y.

I build AI agents and document workflows for research-heavy business problems, with traceable sources, structured outputs, and practical quality checks. I combine more than five years of professional business research with hands-on AI product development. I can own a focused engagement from defining the question and quality criteria to building, testing, and documenting the deliverable. How I can help: • AI agents and document workflows: document Q&A, research synthesis, information extraction, and source-linked reporting. I use Python, LangGraph, and Pydantic AI where they fit the task. • LLM evaluation and diagnosis: expert-informed rubrics, test cases, error analysis, LLM judge calibration, and model quality/cost comparisons. Deliverables can include an evaluation dataset, a diagnostic report, or a scoped improvement prototype. • Business and market research: competitor analysis, market research, business model analysis, and due diligence. Reports distinguish evidence, interpretation, and unresolved questions. Relevant work: • Delivered an AI Q&A assistant for Shanghai Jahwa's finance department and received formal company commendation. • Built Sustainability Workbench, an open-source ESG drafting application with source tracing, human revision, deterministic checks, and Word export. At GreenStone ESG, I also developed LLM evaluations with domain experts. • Built Writfind, an AI-assisted adverse-media research workflow with source-linked findings, citation checks, and structured company risk reports. • Led more than 10 collaborative research projects at IMA and authored hundreds of pre-IPO due diligence reports at Mergermarket. Co-authored a study using GPT-4 to analyze 5,000 consumer reviews, with a 100-review comparison against independent expert ratings. Education: CEIBS MBA, HEC Paris exchange, and MSc Global Prosperity from University College London.