research

own-baseline validation for single-cell potency scores

independent, single-author, open source. may 2026 - present.

the standard way the field validates a potency score cannot separate biological signal from a sequencing-depth confound. i replaced it with a conditional test: does the score order cells better than its own low-order statistic already does?

calibrated on 39,505 cells across twelve microscopy-staged timepoints, where the ordering is fixed before the cells are processed. audited thirteen published scores: of the seven testable, five survive residualization and two do not.

code, mit licence: github.com/IskakovDamir/potency-ownbaseline

status: in preparation

the rise of the ai-augmented solo-founder

redefining team structures in early-stage startups. jun - dec 2025. doi.org/10.5281/zenodo.18242188

interviewed 100+ founders from y combinator, startx, and unicorn companies. co-authored with aldamzhar kiikbayev, guided by prof. poh kam wong (nus business school) and geoffrey moore.