Causal effect of ranking position on clicks
Causal Inference
Recommendations
AIPW
Causal ML
A notebook-driven causal inference project estimating whether top-3 recommendation exposure increases click probability in MIND impression logs.
Causal inference and AI systems for decisions that matter
Research-grade methods, production-aware AI systems, and decision-focused communication.
Portfolio homepage for an Assistant Professor of Data Science focused on causal inference, trustworthy AI systems, and industry decision-making.
Causal Inference and AI Systems for Industry Decisions
I am an Assistant Professor of Data Science building a public portfolio around two connected strengths: credible causal evidence and trustworthy LLM-powered AI systems for real decision workflows.
Question -> Evidence -> AI System -> Decision
Causal Design
→
AI Workflow
Assumptions
Evaluation
Monitoring
From messy data and model outputs to reliable action.
Causal Inference & Experimentation
Evidence for intervention questions: experiments, quasi-experiments, causal ML, policy evaluation, and sensitivity analysis.
AI Systems & LLMOps
RAG, agents, multimodal retrieval, evaluation, hallucination checks, drift monitoring, and lifecycle management.
Causal Methods
DAGs, experiments, DiD, IV, RD, synthetic control, doubly robust estimation, causal ML.
AI Systems
RAG, agents, multi-agent orchestration, vector search, multimodal embeddings, LLM evaluation.
Output
Reusable notebooks, teaching notes, package tutorials, system designs, and decision-ready summaries.
Featured Work
No matching items
Knowledge Hub
Lecture Notes
Short, rigorous explainers on assumptions, identification, estimators, and sensitivity analysis.
Package Tutorials
Practical walkthroughs for the Python tools used in causal inference workflows.
Writing
Blog-style essays that translate methods into business and policy judgment.
Current Roadmap
- Publish three polished industry case studies with notebook-backed evidence.
- Add AI system case studies for RAG, agents, multimodal retrieval, and evaluation.
- Build a lecture-note sequence from causal questions to sensitivity analysis.
- Add package tutorials for causal inference and AI engineering workflows.