Building production-ready ML systems for genomics, clinical decision-making, and discovery.
Adaptive data systems that unify clinical, genomic, and operational data while supporting traceability, versioning, and auditability. Built to power AI-driven reasoning and decision workflows—not just analytics.
AI systems that go beyond prediction. We design agentic workflows that combine models, tools, and domain logic to reason, plan, and adapt over time. Inspired by expert systems, built for high-stakes biomedical and clinical deployment.
Cloud-native AI platforms designed for reliability, governance, and continuous improvement. We enable secure orchestration of AI services, outcome-driven learning, and clear separation between experimentation and production. GCP-first.
Built scalable ML pipelines leveraging transformer-based protein language models for structure-aware representation learning and exploratory biological analysis. Implemented distributed training, model evaluation, and deployment workflows on GCP using modern MLOps practices.
Adapted Geneformer (transformer pre-trained on single-cell RNA-seq) for bulk tumor survival prediction using TCGA data. Demonstrated strong correlation with clinical outcomes and improved patient stratification across tumor stages.
Pre-trained transformer model for multi-class cancer tissue-of-origin prediction across diverse tumor types, including metastatic disease. Supports diagnostic precision and downstream treatment decisions.
Graph-based learning and retrieval-augmented generation for knowledge extraction, search, and summarization across structured and unstructured biomedical data.
Production ML platform deployed on Vertex AI for personalized rheumatoid arthritis therapy. Integrated EMR, claims, genomic, and clinical trial data in a Medicare-covered AI product.
Applied ML research in GC-MS metabolomics for tuberculosis diagnosis and disease severity monitoring through multi-institution clinical collaboration.
Short engagements to evaluate data readiness, model feasibility, and infrastructure design.
Design and implementation of ML systems, data pipelines, and cloud infrastructure, from proof-of-concept through deployment.
Ongoing technical guidance on ML strategy, system design, and scaling in regulated environments.