My work sits at the intersection of protein engineering, immunology, and computation — all directed toward one question: can we teach the immune system to choose tolerance?
Efferocytosis-Mediated Tolerance
Efferocytosis — the process by which phagocytes clear apoptotic cells — naturally promotes immunological tolerance. My central research hypothesis is that this pathway can be co-opted to induce antigen-specific tolerance by engineering recombinant proteins that simultaneously engage efferocytic receptors and present disease-relevant antigens.
This approach has the potential to treat autoimmune diseases and allergies without the broad immunosuppression of current therapies — replacing blanket immune dampening with precision re-education.
Modern immunology generates high-dimensional data that demands new analytical and visual approaches. I work with high-parameter spectral flow cytometry (25-color) and build analysis and visualization workflows in R and Python. I also lead the single-cell multiome and spatial-transcriptomics analysis arm of a Colton Center for Autoimmunity grant on human rheumatoid arthritis.
This work includes adapting visualizations techniques from transcriptomics — such as heat map generation and hierarchical clustering — to flow cytometry data, enabling richer insights into how therapeutics reshape immune cell populations.
Cytokines are powerful immunological signals, but their therapeutic use is limited by short half-lives, systemic toxicity, and off-target effects. Through protein engineering — including albumin fusions, prodrug strategies, and tissue-targeting moieties — we design cytokines that act where and when they're needed.
This collaborative work spans autoimmune diseases (EAE, rheumatoid arthritis), cardiovascular inflammation (atherosclerosis), and wound healing — demonstrating that engineered cytokines can suppress pathological inflammation while preserving protective immunity.
I treat modern AI as lab infrastructure. Using Claude Code, I build custom tools that remove friction from experimental work — from correcting file-corruption bugs that cost days of reanalysis to standing up reproducible computational pipelines across two campuses' HPC systems.
Most recently, I operationalized a released molecular-recognition foundation model into a working inference pipeline for in-silico protein–protein-interface screening, gated by structural cofolding — bringing a published model that didn't run out of the box into practical, validated use in the lab.