Jiaqi Zhang to Join Columbia University as Assistant Professor in Computer Science
Jiaqi Zhang announced she will join the Computer Science Department at Columbia University as an Assistant Professor in January 2027, focusing her lab on causal inference, artificial intelligence, experimental design, and biomedical applications, and will begin recruiting PhD students and postdocs.
She obtained her PhD from MIT EECS under the supervision of Caroline Uhler and completed her undergraduate studies in mathematics at Peking University. She has interned at ByteDance, Microsoft Research, and Apple. Her research focuses on identifying causal relationships from data, extrapolating intervention effects, and selecting the most informative interventions, serving cell biology and biomedical discoveries. Relevant work includes causal decoupling identifiability, active learning for intervention design, and single-cell perturbation modeling, and she has been involved in organizing the CLeaR conference and the NeurIPS causal science workshop.
The recruitment window is aligned with this fall's PhD applications in computer science. The lab will be located in New York, integrating statistical causality with wet lab design into the same group. Columbia's Computer Science Department aims to incorporate causal and biomedical computation into its faculty roster, connecting with the perturbation prediction networks around MIT and the Broad Institute.
In market mechanisms, this represents the academic establishment's pricing of talent in causal and perturbation prediction. Beneficiaries include departments looking to establish cross-disciplinary labs and biomedical labs needing intervention design tools; the pressure point is private labs in similar fields still competing for the same pool of PhD students. Funding comes from faculty startup packages and subsequent federal and foundation grants, without going through product releases.
She will start in January 2027. Publications and conference organization are existing outputs, while the lab size will depend on the first cohort of admissions.
Source: Public information
ABAB AI Insight
Causal inference moves from theoretical identifiability to single-cell perturbations, with this faculty position anchoring that line in an Ivy League computer science department rather than being confined to statistics or medical schools. Uhler's training focuses on "which intervention is most worthwhile," with pharmaceuticals and cell atlases following this logic for computational purchases. New York offers proximity to hospitals and genomic centers, as well as a source of PhD students. There is no issuance of currency or financing; the startup package represents this round of capital expenditure.
The talent market positions causal and biomedical as a scarce combination: there is a high supply of pure large model positions, but a low supply of those who can translate intervention design into algorithms. Internships at ByteDance and Apple indicate that the industry seeks similar tools, but ultimately, the establishment returns to academia, as wet lab collaboration fits better with academic timelines.
This contrasts with various schools competing for AI + life sciences assistant professors and the perturbation prediction competition around Broad. The industry is moving from correlational prediction to actionable prediction: whoever can tell the experiment to target which gene first will enter the experimental budget.
This reflects the projection of technological substitution onto academic establishments: random screening is replaced by experimental designs that maximize information. The mechanism is that faculty positions establish methods, while PhD students turn those methods into reproducible pipelines.
ABAB News · Cognitive Laws
Being able to decide which intervention to perform first is scarcer than re-fitting a correlation.
In biomedical computation faculty positions, what is purchased is the proximity of the lab, not the scale of parameters.
Modeling predictions of cellular responses determines who can translate predictions into the petri dish.