I work on transcriptomics, functional genomics, and single-cell analysis. My day-to-day covers bulk RNA-seq and microarray analysis, differential expression (DESeq2, limma), functional enrichment (GO, KEGG, g:Profiler), and protein–protein interaction networks (STRING), together with single-cell and single-nucleus RNA-seq (sc/snRNA-seq) run on HPC systems, including work on Huntington's disease model organisms. A recurring theme is non-model organisms (crops, filamentous fungi, and yeasts), such as Fusarium graminearum in wheat and barley, that lack ready-made Bioconductor annotation, so I spend most of my time building reproducible, portable pipelines in R and Python.

Lately I turn those methods into finished tools — local-first, reproducible, and provenance-sealed: a no-code desktop app for reproducible RNA-seq and microarray analysis (BulkSeq Studio), a desktop tool for transposable-element annotation and TE-aware PCR primer design (TEagle), a leakage-safe multi-omics data-audit and fusion-benchmarking CLI with a clinical dashboard and deep-learning fusion (omicau), and a local-first knowledge tool for scientists and coders (Arf). I care about how software looks and feels as much as how it runs — design, typography, and visual craft are a standing artistic curiosity for me, most visible in Arf's interface.

I'm also deeply interested in epistemology and in the methodology and history of science — how knowledge is justified and how methods come to be trusted. It's the same instinct that pulls me toward reproducibility and honest evidence in the tools I build.

Languages
Bulk RNA-seq
Single-cell
Multi-omics ML
Genomics
Resources
Reproducibility
Desktop / annotation

Arf — a local-first second brain for scientists and coders: plain-Markdown notes with [[wikilinks]], LaTeX and syntax-highlighted code, a knowledge graph, and on-device embeddings (MiniLM) that surface related but unlinked notes. Native desktop app for Windows, macOS, and Linux. Download.

Litehouse (live) — a browser-only scholarly literature-review tool: it retrieves works from open APIs (OpenAlex, Crossref, Europe PMC, DataCite), then writes an evidence-locked, cited synthesis with a local WebGPU model (Qwen3), attaches SHA-256 integrity receipts, and exports a classic Computer Modern LaTeX report. No server and no install — retrieval, the model, and rendering all run in the browser.

BulkSeq Studio — a cross-platform, no-code desktop app for reproducible bulk RNA-seq and microarray analysis: STAR / HISAT2 / Salmon alignment, DESeq2 / limma, GO / KEGG / g:Profiler enrichment, and STRING interaction networks, with first-class support for crops and fungi that lack a Bioconductor OrgDb.

TEagle — a native desktop app for transposable-element annotation and TE-aware PCR primer design: evidence-traceable structural and protein-domain (HMMER) classification, an interactive genome viewer, Primer3 design checked by pair-aware in-silico PCR, and a local whole-genome off-target scan (RepeatMasker / Dfam and minimap2 run through a managed WSL backend) — with every result sealed by content-addressed provenance (database and tool versions plus checksums) so it reproduces exactly. Windows, no command line. Download.

omicau — a reproducible, leakage-safe multi-omics data-audit CLI: format-agnostic ingestion and alignment, SHA-256 data provenance, missingness-bias and batch-effect diagnostics, group-aware cross-validated classical and PyTorch masked-pooling fusion benchmarks with leakage-safe feature attribution, and a dual clinical/research dashboard. Built for the Build with Claude: Life Sciences hackathon (Anthropic × Gladstone Institutes, Jul 7–13 2026).

geneRelate / pomRelate — a pair of static, client-side tools for cross-species gene analysis that run entirely in the browser with no backend: ortholog mapping, STRING protein–protein interaction networks, GO / KEGG enrichment, and per-gene phylogenetic trees. geneRelate (demo) covers 20 Fusarium species; pomRelate (demo) centers on Schizosaccharomyces pombe across model organisms. Built on STRING v12.0, KEGG, and eggNOG v7.