Research
My research asks a practical question: how should organizations adopt AI systems that are powerful, cheap, and unreliable at the same time? I work on this from three directions: the economics of AI adoption, the governance of autonomous agents, and applied systems built with industry partners.
Economics of AI adoption
Innovation costsDraft
An essay on accelerable versus scalable innovation, using lessons learned from the history of technology.
Applied systems with industry partners
Better Business Bureau of AtlantaActive
Over 200,000 consumer complaints. Two tracks: a BBB analytics dashboard for cohort and archetype analysis and demand sensing from complaint streams. And, ChatBBB, conversational agents for complaint intake and resolution that are aware of emotion, complexity, history, and when to escalate. A journal special-issue submission is planned.
Causal inference with LLMsOngoing
IV Finder: a pipeline that reads academic papers to propose candidate instrumental variables and checks relevance and exclusion arguments. Demonstrated on the ESG-to-profitability question.
Earlier work
- Prompt2Cypher. Natural-language to Cypher query generation for biomedical knowledge graphs (ProKinO, ICKG).
- Supply chain knowledge graph and the CogEval benchmark. Building and evaluating a supply chain knowledge graph, with the CogEval benchmark for cognitive-style evaluation of language models.
- Kinase SNP analysis. Explainable machine learning for kinase-substrate interactions and phosphorylation prediction (Phosformer), and annotation of understudied "dark" kinases.
Publications and essays
The practitioner pieces are where I test whether an idea is useful; the papers are where I test whether it is true.
Articles and essays
The cost of innovation: accelerable versus scalableDraft
Human Genome Project versus Celera, and what it means for AI.
Do LLMs reduce criticism?Idea
Notes on how AI assistants change the culture of feedback. And, how it affects downstream systems.
Academic publications
Prompt2CypherUnder revision
Natural-language querying of biomedical knowledge graphs with LLMs (preprint: “Task Splitting and Prompt Engineering for Cypher Query Generation”, bioRxiv 2025).
Socratic Method Revisited: Human-AI Dialogue for Knowledge Creation
Guided-conversation AI as a method for knowledge creation.
CLASPP: A unified model for predicting post-translational modifications
A single model for predicting multiple post-translational modifications.
Identification and classification of ion channels across the tree of life
Sequence-based identification and classification of ion channels.
Flood of techniques and drought of theories: emotion mining in disasters
A review of emotion mining in disaster contexts and its thin theoretical base.
Using explainable machine learning to uncover the kinase-substrate interaction landscape
Explainable models of which kinases phosphorylate which substrates.
Phosformer: an explainable transformer model for protein kinase-specific phosphorylation predictions
A transformer for kinase-specific phosphorylation site prediction.
Dark kinase annotation, mining, and visualization using the Protein Kinase Ontology
Ontology-driven annotation of understudied kinases.
The cost of looking natural: why the no-makeup movement may fail
Consumer responses to natural-looking versus made-up appearance.
GTXplorer: a portal to navigate and visualize the evolutionary information encoded in fold A glycosyltransferases
A web portal for evolutionary analysis of glycosyltransferases.
Socially aware multimodal deep neural networks for fake news classification
Combining text, images, and social context to classify fake news.
Personalized feedback emails: a case study on online introductory computer science courses
Automated personalized feedback for students in online CS courses.
Curtailing fake news propagation with psychographics
Using psychographic profiles to understand the spread of fake news.