Right now, Michael Vidne is building quietly. His latest venture, Baobab Therapeutics, sits in stealth mode at the place he keeps returning to: where AI, data, and medicine collide. No press kit, no banner asset, just the same instinct that has run through his whole career - that complex systems leave readable signals, if you build the right model to catch them.
That instinct started somewhere unexpected. Before cancer, before consulting decks, before a single board meeting, Vidne studied vision. His doctorate at Columbia's Center for Theoretical Neuroscience was a deep dive into how a densely sampled network of more than 250 retinal neurons fire together - how a crowd of cells, awash in shared noise, still manages to encode a coherent picture of the world. It is an oddly fitting origin for a man who now spends his days asking a structurally identical question about tumors: what is this whole network of cells actually doing, all at once?
The leap from the eye to oncology was not a pivot so much as a translation. Same toolkit, different cells.
We monitor the activity of all the signaling pathways that originate with the patient's gene.
A founder who keeps choosing the hard, quantitative middle
Baobab Therapeutics is the current chapter, and it is deliberately under-lit. What is public is the thesis: Vidne describes himself as working at the intersection of AI, data, and healthcare, and he posts about the responsible use of large language models in drug development - the unglamorous question of how you let a model help design a medicine without letting it hallucinate one. It is a founder's bet that the next edge in biotech is less about a single molecule and more about the machinery that decides which molecule, for which patient, when.
This is consistent with everything he did before. At NovellusDx, the platform was functional rather than purely genomic: it effectively cloned a patient's cancer biopsy onto a chip, then watched the tumor's signaling pathways light up - or stay dark - in response to candidate drugs over a span of five to seven days. The genome tells you what mutations exist. Vidne's instrument was built to tell you what those mutations are doing.
His complaint with sequence-only medicine was blunt: it misses driver mutations, and it infers behavior it never actually measures. His fix was to measure.
The Through-Line
- Retina: model how a network of neurons encodes a signal.
- Tumor: model how a network of pathways responds to a drug.
- AI era: model how data and language models can decide what to test next.
Different domains. One stubborn question: read the signal, then act on it.
True personalized medicine means characterizing the whole tumor and testing the drugs against it - not inferring from the sequence alone.
How a vision scientist ended up in cancer drug development
Education
Columbia University
Ph.D., Applied Mathematics - Center for Theoretical Neuroscience (2004-2011). Doctoral work on statistical models of the visual system and the network activity of retinal ganglion cells.
The McKinsey Detour
A West Coast consulting stint that bridges the gap on his resume between neurons and oncology - and explains why the scientist could later run the boardroom.
From a Jerusalem lab to Market Street, Philadelphia
NovellusDx
Founded 2011 in Jerusalem. A functional, live-cell platform that cloned cancer biopsies onto a chip to read which mutations were truly driving the disease - and which drugs could switch them off.
Fore Biotherapeutics
The U.S. operating company born from NovellusDx's pivot. Strategy: match clinical-stage compounds to the precise patient populations their data fits. Lead asset: the BRAF inhibitor plixorafenib.
Baobab Therapeutics
Vidne's current, stealth-mode venture. Public thesis only: AI, data, and healthcare - and the responsible use of large models in how medicines get developed.
The Backers
The Fore lineage drew OrbiMed, Novartis Venture Fund, and Wellington Management - a $57M Series C in 2020, with total funding across the lineage approaching $204.5M.
Things you would not guess from the org chart
His scientific roots are in the eye - modeling how retinal neurons collectively encode what we see.
He once presented a neuromorphic technologies program directly to the President of Israel.
He went from analyzing how neurons signal to analyzing how cancer cells signal - the structure of the problem barely changed.
Theoretical neuroscience, management consulting, cancer drug development - one modeling toolkit across all three.
The paper trail
Profile assembled from public sources: company filings, press coverage, and Vidne's own published scientific work. Where the record was thin, we left it thin.