# about

# published research projects

bioRxiv 2026 · first author

every lesion, every timepoint

i clustered 1,193 lesions from 110 animals into severity classes, then built sequential random-forest models that predict how the severe ones respond to 10 regimens, and explained the predictions with SHAP.

  • 1,193 lesions
  • 22 regimens
  • 6 timepoints

[paper][code][full view]

Cell Systems 2025 · co-author

two modalities, one latent space

i am responsible for DECIPHAER's modeling development: the autoencoders, the adversarial training that merges transcriptional and morphological data into one latent space, and the work that continues to build on it.

  • 2 encoders
  • 128 latent dims
  • 1 adversary

[paper][code][full view]

Scientific Reports 2026 · co-author

how one-sided is the signal?

i wrote the image analysis that segments each nucleus, collects the perinuclear signal around it, and reduces it to one number: the intensity-weighted circular resultant length, R.

  • 545 nuclei
  • 170 images
  • R ∈ [0, 1] one score

[paper][code][full view]

# other projects

# papers

    # contact

    the reliable way to reach me is email or linkedin, everything else:

    contact form →