Three different kinds of consulting, listed most to least technical. The common thread is what I enjoy about the work: someone brings a problem stated in their own terms, and the first task is working out what it actually is before any code gets written.
Computational biology
Most of this is independent work, and most of it arrives by word of mouth: bioinformatics and computational-biology projects I take on directly for researchers, outside any lab I belong to. One such collaboration earned me co-first authorship on a study of mycorrhiza-enhanced salt tolerance in rice (Frontiers in Plant Science, 2022), where I did the comparative transcriptome analysis as the outside analyst rather than a lab member. Increasingly, colleagues also come to me to put agentic AI and LLM tooling to work in their own research.
What I can help with maps closely to what I’ve spent six years doing:
- RNA-seq, end to end. Raw FASTQ through differential expression, including the unglamorous parts: sample QC, batch structure, and deciding what a comparison can and cannot support. Built on STAR, HISAT2, DESeq2, and edgeR.
- Reproducible pipelines on HPC. Snakemake and SLURM workflows that a second person can rerun and get the same answer. I have reprocessed large-scale community datasets this way and written the templates I use for GPU and HPC jobs.
- Machine learning on genomic data, validated honestly. Most of my value here is negative results: catching leakage before it becomes a finding. Chromosome-held-out cross-validation, permuted-label controls, and matched random baselines are the default, not the extra credit.
- Foundation model evaluation. Whether a pretrained DNA or RNA model helps on your organism and your task, which one to reach for, and how to pool its embeddings. I have evaluated seven of them on a non-model plant genome and found ways to make six fail under biologically valid edits.
- Comparative genomics and phylogenetics. Gene family identification, orthology, duplication analysis, and genome-wide annotation.
I would sooner tell you a question is not answerable with the data at hand than produce a figure that implies it is. Much of my dissertation has gone into building ground-truth reference sets to check whether automated pipelines find real biology or artifacts, and that habit carries over.
This fall I am also contributing part-time to the CBCG (Computational Biology Consulting Group), an initiative at the UGA Institute of Bioinformatics led by Dr. Casey Bergman that gives labs with a computational problem somewhere to turn. My work there is developing a 16S rRNA metagenomics pipeline — DADA2 and QIIME2 with the SILVA reference database, alpha- and beta-diversity analysis, and differential-abundance testing with MaAsLin2 — applied to a mouse gut microbiome dataset. A CBCG assistantship supported me this semester, for which I am grateful.
Technology — Project TYRA
Technology Consultant and IT Group Convener, Taiwanese Young Researchers Association. April 2022 – present.
A volunteer role that became the longest-running thing I have built. The organization needed to match mentors to mentees at a scale their spreadsheet could not handle, so I architected the platform behind it.
- Grew a mentor–mentee matching program from 274 to 700+ annual participants across 4 cohorts, facilitating 1,876 researcher connections
- The organization later partnered with Fulbright Taiwan and government education agencies
- Deployed an automated event pipeline (Eventbrite API plus scheduling automation) for the weekly seminar series
The consulting part was not the code. It was working out that the real constraint was reviewer attention, not matching accuracy, and building for that instead. I also automated transcription of the weekly seminars with Whisper (TYRA_transcript) so the talks stayed searchable. Write-up on the project page.
Management consulting — in training
Director of Marketing and Communication, PhD Consulting Club, University of Georgia. July – December 2025.
I am preparing for management consulting alongside the technical work, and I want to be precise about the stage: this is training and club leadership, not client engagements.
- Tripled club membership across multiple departments
- Ran weekly structured case practice sessions
- Working through case frameworks and interview preparation
The overlap with the technical work is real. Scoping an ambiguous question, deciding what evidence would actually settle it, and telling someone their data does not support the conclusion they want are the same skills in both settings.
I like this kind of problem and am glad to be approached with one. If you’re at UGA, computational-biology work runs through the Institute of Bioinformatics — that’s the way to reach the CBCG. For anything else, or to talk a problem through before it’s fully formed, I’m reachable by email.