Co-founder & CTO of Via Scientific, Associate Professor at UMass Chan Medical School. Ten-plus years building multi-omics and AI systems used by thousands of scientists across academia and biopharma.
I lead the technical strategy for Via Foundry, a multi-omics and AI platform commercialized from UMass Chan and used by biotech, pharma, and academic research teams. The work covers modern ML, cloud data engineering, and biomedical science, and all three have to hold up under real use.
I've built or co-authored tools cited tens of thousands of times (I-TASSER, DolphinNext, DEBrowser, GUIDEseq) and published in Nature, Cell, Science, Nature Protocols, and Nature Communications. The goal of everything I ship is the same: to make it obvious whether the data in front of you is telling you something true, and whether the next decision is a good one.
I've sat on both sides of the table: founder and CTO of a venture-backed platform company, and faculty at UMass Chan with publications in Nature, Cell, and Science.
I own the path from feasibility and MVP through production scaling, MLOps, and monitoring. Standards and reference architectures that hold up in practice, not just in slide decks.
Ten years building reproducible, version-controlled analytics in biomedical settings where data provenance, validation, and auditability are required, not optional. Familiar with GxP, 21 CFR Part 11, and the EU AI Act.
Recruited and mentored multidisciplinary teams of data scientists, AI/ML engineers, software engineers, and computational biologists. Day-to-day collaboration with scientists, clinicians, product, legal, and investors.
Metrics, drift monitoring, and the kind of stakeholder communication that tells you when the model is working and when it isn't. I'd rather ship a smaller thing that's honest about its limits than a bigger one that isn't.
Owning technical strategy, architecture, and engineering standards for Via Foundry, a multi-omics & AI analytics platform exclusively licensed from UMass Chan and used by biotech and pharma research teams. Venture-backed, with advisors that include Melissa J. Moore (former CSO of Moderna), Rob Hickey (former EVP Engineering, DataRobot), and Shah Nawaz (former VP Digital Transformation, Regeneron).
Faculty appointment in the Program in Molecular Medicine and the Department of Genomics & Computational Biology. Research focused on scalable, reproducible multi-omics analytics, RNA therapeutics, and applied ML for biomedical data.
Promoted into the faculty track to lead the scale-up of the Bioinformatics Core. Designed and built the NGS analytics stack that ultimately became Via Foundry; standardized RNA-Seq, scRNA-Seq, ChIP-Seq, ATAC-Seq, and GATK-based variant-calling pipelines; and delivered enterprise-grade training programs across the institution.
Senior engineer on the Bioinformatics Core. Designed and productionized reusable, robust pipelines for NGS, microarray, genomics, proteomics, and chemogenomics data — supporting hundreds of concurrent users on HPC clusters. Installed and maintained Galaxy, GenePattern, and Genome Space for the institution; built custom databases, portals, and visualization apps where off-the-shelf tools fell short; led institution-facing training workshops.
Worked in the Melissa J. Moore and Phillip D. Zamore labs on post-transcriptional gene regulation via RNA mechanisms. Built bioinformatics software and visualizations for SOLEXA and SOLiD NGS platforms, peak-calling and denoising algorithms (the foundation for ASPeak), machine-learning pattern-search methods, custom scientist-facing interfaces, and UCSC Genome Browser mirrors for institutional use.
Co-developed I-TASSER, the unified platform for automated protein structure and function prediction. Co-author on the Nature Protocols 2010 method paper — now cited 7,800+ times — and contributor to subsequent I-TASSER methodology development that remains a reference tool in computational structural biology.
Dissertation on graph-theoretic discrimination of native protein folds. MS in Systems Analysis and BS in Mathematics Engineering, both from Istanbul Technical University.
Each of these is open, documented, and in daily use by researchers outside the lab that built it. Together they've become common infrastructure for NGS and protein analysis.
Multi-omics and AI analytics platform. Drag-and-drop pipeline builder, versioned reproducibility, cloud-native execution. Licensed from UMass Chan and used by biotech and pharma research teams.
Distributed data processing platform for high-throughput genomics. GUI-driven Nextflow pipeline authoring with institutional reproducibility and HPC/cloud execution. BMC Genomics, 2020.
Interactive differential expression analysis and visualization tool for count data. First author; 287+ citations, distributed via Bioconductor. BMC Genomics, 2019.
Unified platform for automated protein structure & function prediction. Co-authored the Nature Protocols (2010) method paper, cited 7,800+ times and still a reference tool in computational structural biology.
Bioconductor package to analyze GUIDE-Seq datasets for CRISPR-Cas nucleases — standard in CRISPR off-target analysis. BMC Genomics, 2017.
Abundance-sensitive peak detection algorithm for RIP-Seq. First-author method paper in Bioinformatics (2013), standard in RNA-protein interaction studies.
Full bibliography on Google Scholar: 35+ peer-reviewed publications, 13,500+ citations, h-index 26. Ten representative ones below, covering AI methods, multi-omics platforms, and biology.
Posts on Medium about reproducibility, pipeline design, scRNA-seq, RNA therapeutics, and the day-to-day of doing science with AI.
Reproducibility and collaboration as survival tools in a field where the methods move faster than the questions.
After writing the same pipeline for the sixth time, the lesson finally stuck. An argument for reusable, versioned analytics.
Why bulk RNA-seq isn't obsolete — and how modern platforms keep extracting new signal from the same modality.
The five places spatial transcriptomics projects go wrong — and how to design the data layer so they don't.
How Via Foundry served as the analytical backbone of a longitudinal autoimmune disease investigation.
Single-cell demands three jobs at once. Why the tooling has to meet scientists where they are.
Reproducibility isn't a report — it's a handoff. A team-scale view of what "reproducible" has to mean.
How spatial transcriptomics, combined with multi-omics infrastructure, is changing the RNA therapeutics pipeline.
Where RNA therapeutics are stuck — and where computational biology is doing the most to unstick them.
Why no credible RNA therapeutic ships without the bioinformatics layer doing heavy lifting upstream.
Open to conversations on leadership roles in data science / AI engineering, advisory work, and research collaborations — especially in pharma, biotech, and translational medicine.