Solhee Tucker
ML / Bioinformatics Engineer
Grounded in real data, not just plausible output, with a solid software engineering background behind it.
An AI agent proposes every analysis decision and backs it with a citation. I review each one, confirm or override it. Fixed demo on two real datasets: colorectal cancer and Parkinson's disease.
Histone marks near a gene's TSS predict how much it's expressed, mostly from one feature. I checked it against four regression models to rule out a modeling artifact.
Pulls NCBI and Ensembl data and grounds the summary in real PubMed citations instead of model recall. Fixed demo from a bioinformatics hackathon.
COVID-19 blood has more activated effector cells and fewer resting naive and memory cells. Built from CELLxGENE Census data into an interactive chart.