Projects

Preterm brain & cerebellar development

Carries out the bioinformatics and computational analysis behind the lab's preterm-cerebellum research on how preterm birth, necrotizing enterocolitis (NEC) and SIDS reshape the developing cerebellum, across two main threads: behavioral phenotyping of 250+ mice (DeepLabCut, B-SOiD) plus statistical and machine-learning models on Erasmus Ladder locomotor-learning data from 160 mice, to detect gait and cerebellar deficits and the effects of physical therapy and enrichment; and longitudinal analysis of multi-domain records from 200+ preterm infants (MRI, cranial ultrasound, Fenton growth charts, neurodevelopmental scores) to characterize NEC trajectories. Also produces the figures and scientific writing.

Relocated to the US in mid-2022 for a spouse's work assignment; used a visa-related career break to re-skill in Python and R and complete an M.S. in Data Analytics (Western Governors University, 2024–2025).

MyVivarium — open-source lab-animal platform

First-authored and built the full stack of MyVivarium, an open-source cloud app for managing animal colonies, developed under Digirobi Solutions in collaboration with the Sathyanesan Lab (Dr. Aaron Sathyanesan, University of Dayton). Published in 2025, it provides collaborative tracking, QR-code cage cards, task reminders, and a low-cost IoT system streaming temperature, humidity, light and activity in real time, deployable for roughly the cost of a website. Deployed at three partner labs (200+ animals); a v2 redesign is now in development.

Neonatal brain-injury modeling

Contributed the mechanistic modeling to a collaborative neonatal-neuroscience study. Built a MATLAB SimBiology kinetic model of calcium/calmodulin–Src-kinase signaling in cortical neurons under hypoxia (piglet model), validated with Sobol and global sensitivity analysis — identifying a post-hypoxia window where modulating Src–NMDAR kinetics (PP2) limits calcium influx and pro-apoptotic signaling.

Cancer-nutrition databases & patent (consulting)

Founded Digirobi Solutions, an independent computational-biology and analytics consultancy, serving as the computational biologist on a client's cancer-nutrition project. Mined open and proprietary sources to build the cancer-genomics, drug, and nutrition databases behind the client's nutrigenomics tool, which links dietary compounds to cancer-relevant pathways, and co-invented the resulting granted US and European patents, whose rights were later acquired by another company. The consultancy also supported non-biomedical clients with marketing analytics and full web and server operations.

Multi-cell immuno-oncology biomarker modeling

Combined single-cell computational models into multi-cell “virtual tissue” predicting the chemokine, cytokine and cellular-biomarker profiles of inflamed tissue and the tumor microenvironment — matching laboratory cultures 75–80%, and personalized with multiple-myeloma cell-line genomics. Proposed as a high-throughput screen for anti-inflammatory and immuno-oncology therapeutics.

BEAT AML drug-response prediction & digital twins

Led the computational modeling (and second author) for the BEAT AML collaboration: built patient-specific “cancer avatars” (digital twins) from each genome and ran virtual drug tests, predicting BET-inhibitor response in 93% of 100 AML patients. Built the data engine behind it: a 500+ cancer cell-line database with IC50 profiles (GDSC, CCLE, COSMIC, NCI-60) plus TCGA/cBioPortal. Also led a 10+ scientist team, setting model-validation standards.

Drug knowledge base & digital drug models

Built a ~200-compound drug knowledge base (targets, mechanism, pharmacokinetics, resistance) and turned each into a simulatable “digital drug” inside the disease models. Mentored across 25+ training sessions and led the claim drafting and prosecution of the team's cancer-combination patent.

Mucosal & innate-immune modeling — most-cited work

Contributed the in-silico dendritic-cell modeling and analysis for antimicrobial-peptide collaborations — predicting, and validating with lab partners, how the defensin DEFB103 and histatin 5 tune inflammation up or down. These remain among the most-cited contributions, together accounting for 100+ citations — the majority of the 150+ total.

Drug-repurposing platform & predictive-modeling methods

Built the methodology that made high-throughput screening on the disease models scalable: a proprietary command language for scripting virtual experiments, executed across a Load Sharing Facility (LSF) compute grid that distributed simulations over networked machines and automatically collected, compared and scored the results. Ran high-throughput virtual drug-efficacy and repurposing screens on this platform — presented as a novel predictive-modeling approach at ACR 2012 and protected by a method patent. The platform was adopted and further developed in-house, becoming part of the company's growth story and its case for venture investment.

Rheumatoid-arthritis model & repurposed combination therapy

Modeled 20+ interconnected signaling pathways across macrophages, T cells and other immune cell types, and led their integration into a rheumatoid-arthritis disease model that grew from four cell types to eight, validated against in-vivo data. Using high-throughput virtual screens, identified two synergistic repurposed-drug combinations advanced for clinical collaboration; the lead combination showed profound impact on disease progression in a murine collagen-induced arthritis model. Protected by two composition-of-matter patents.

Immune-signaling pathway modeling

First project: mechanistic ODE models of interleukin and TNF signaling in macrophages spanning 20+ interconnected pathways with perturbation-based validation, extended into an arachidonic-acid / EET model showing how a soluble epoxide hydrolase inhibitor calms NF-κB-driven inflammation. This early work helped shape the perturbation-and-validation approach later used across the company's computational-biology platform.