Staff Data Scientist | AI Engineer | Agentic Systems Builder
I build practical AI systems that combine data science, automation, product thinking, and reliable engineering. My work focuses on turning messy business questions into usable tools: agents, audit workflows, ranking/evaluation systems, and decision-support products.
- Agentic AI systems: task-specific agents, tool-using workflows, evaluator loops, and human-in-the-loop automations.
- Applied data science: scoring models, diagnostics, decision frameworks, experimentation, and analytics workflows.
- AI visibility and retrieval tools: technical audits for how websites and brands are discovered, parsed, cited, and recommended by AI answer engines.
- Risk and quality evaluation: structured checks for AI outputs, content quality, reliability, and operational readiness.
| Project | What it shows |
|---|---|
| agentic-ai-risk-auditor | AI governance mindset, risk evaluation, structured diagnostics, and practical safety workflows. |
| daily-news-agent | Agent automation, information filtering, scheduled workflows, and decision-support summaries. |
| SmartReview | Applied AI product thinking, review workflows, feedback analysis, and TypeScript implementation. |
| luna-arts-studio.io | Full-stack product execution with TypeScript, frontend systems, and deployed user experience. |
| OpenClaw-Medical-Skills | Open-source AI skills ecosystem experience and domain-specific agent workflows. |
AI and ML: LLM workflows, RAG patterns, evaluation, prompt systems, agent orchestration, ranking and scoring logic, experimentation
Data: Python, pandas, analytics pipelines, feature design, metric definition, model evaluation, reporting automation
Engineering: TypeScript, FastAPI-style services, CLI tools, GitHub Actions, testing, documentation, deployable prototypes
Product: translating ambiguous goals into scoped tools, clear UX, practical adoption paths, and measurable outcomes
I like building systems that are useful enough to run repeatedly, not just demos that look good once. My default style is:
- define the decision the system should improve;
- make the data, assumptions, and scoring visible;
- ship a small working version;
- add evaluation so quality can improve over time;
- document the workflow so someone else can use it.
- Production-ready AI agents and workflow automation
- Evaluation frameworks for LLM and agent outputs
- AI search, answer-engine visibility, and citation readiness
- Data products for marketing, operations, and decision intelligence
I am especially interested in Staff Data Scientist, AI Engineer, Applied Scientist, and AI Product Engineer roles where the job is to build reliable AI systems with real users, clear metrics, and business impact.
Email: cathyismee@gmail.com