I build evaluation-driven AI systems, clinical NLP workflows, and production-minded ML pipelines with emphasis on reliability, grounding, and measurable signal over polished demos.
I bring a research-trained approach to production AI: careful problem definition, disciplined evaluation, and systems that stay understandable when the work gets messy.
At Carta Healthcare, I built clinical ETL and NLP pipelines with HL7/FHIR-aligned workflows and LLM-assisted extraction. At TripleTen, I mentor end-to-end machine learning projects with emphasis on evaluation rigor, failure analysis, and production-minded system design.
5 peer-reviewed publications across nonlinear dynamics and complex systems. View publications
Case studies in evaluation-driven AI systems, retrieval and grounding, and clinical NLP.
Standalone flagship build · AI Systems Builder
Built an evaluation-first RAG system for dense technical documents, using a labeled benchmark to separate retrieval quality from grounded answer quality before optimizing generation.
Essays on machine learning practice, systems thinking, scientific reasoning, and technical judgment.
In production ML systems, the hardest part is often not the algorithm. It is making data, labels, and evaluation trustworthy enough for the model to matter.
Scientific training strengthens applied AI because it builds the habits that reliable ML systems depend on: questioning assumptions, isolating variables, and taking evidence seriously.
I am especially open to machine learning, AI systems, and research-informed product work where rigor, reliability, and communication all matter.