AI for biotech, verified by someone who has run the assay.
I'm Madeline Valentin. I help biotech and life-science teams decide where AI belongs in their work, and make sure it does what it's supposed to once it's there.
The work
Two questions sit underneath almost every conversation about AI in a lab or a research operation. Which parts of the workflow are a good fit for it, and which are going to produce confident, wrong answers that nobody catches until they cost something? And once a system is in place, how do you know it holds up on your science and not just on a demo?
I work on both: mapping where AI should and shouldn't be used, then assessing how a system behaves against real scientific questions and setting up the checks that catch it when it drifts.
Most people assessing AI systems have never run an assay, and the scientists who could catch the errors usually aren't in the room when the system gets tested. I've worked on both sides.
How it starts
- A 30-minute call. You describe where AI sits in your work, or where you're thinking of putting it. No prep, no pitch.
- A one-week diagnostic on your own material: real outputs from the system you're using, or the questions you'd want it to handle. You get a written read on where it's solid, where it breaks, and what to fix or fence off first.
- If it makes sense, we scope something bigger from there.
Background
M.S. in Biomedical Engineering, University of South Dakota. My lab work contributed to a peer-reviewed 2022 study on GM1 ganglioside's therapeutic effects in Huntington's Disease model cells, where I validated assay findings and supported the data that carried the study to publication. Selected for USD's Neuroscience, Nanotechnology, and Network (N3) Fellowship, with a grounding in intellectual property and responsible research.
Talk to me
If you're bringing AI into research, clinical, or regulatory work and want a straight read on where it fits and whether it's holding up, the fastest way to find out if I'm useful to you is a short call.