Portrait of Jamila Taaki

Jamila Taaki

Statistical inference & computational methods for exoplanets

I'm a Schmidt Postdoctoral Fellow at the Michigan Institute for Data & AI in Society (University of Michigan), mentored by Professor Lia Corrales (Astronomy) and Professor Alfred Hero (EECS). I completed my PhD in Electrical and Computer Engineering at the University of Illinois Urbana–Champaign, advised by Professor Farzad Kamalabadi and Professor Athol Kemball. Before Illinois, I completed an MSc in Astrophysics at Royal Holloway, University of London.

Research interests: detection theory for transit photometry, sub-µas astrometry for stellar-surface mapping, and starshade high-contrast imaging.

Download CV (PDF) Research Statement (PDF)

Publications

Preprints / In review

ORCID

Projects / Software

Astrometric jitter visualizer showing starspot and planet signal

Astrometric Jitter Visualizer

Astrometric jitter from starspots rotating in and out of view may limit our ability to find and measure Earth-like planets orbiting in the habitable zone of Sun-like stars. Visualize how astrometric jitter contaminates true stellar reflex motion from an orbiting exoplanet.

Visualizer · arXiv

Starshade pupil and PSF thumbnail

PyStarshade

Python library for Fresnel diffraction simulations of starshades with realistic telescope apertures, using Bluestein FFTs to move between varying spatial scales at optical planes and resolve high-resolution PSFs.

Code · JOSS paper

Transit light curve still from exoplanet bot

Exoplanet of the Day

Ode to the transit detection technique. Twitter/X bot that posts an animated transit light curve each day, using a real Kepler detection and light curve. Visualizes the transit to scale.

Bot · Code

Raw and detrended light curve panels

spatial-detrend

Low-rank, spatial (sensor) detrending for missions like Kepler and TESS. Removing structured detector systematics with a total variation prior while preserving astrophysical variability.

Code

GPU vs CPU performance plot thumbnail

CUDA Transit Detection

Prototype GPU kernels for fast Bayesian transit likelihood ratio tests with full rank covariance models that account for both instrument systematics and stellar variability.

Code

TESS Joint Bayesian Transit Search

Large-scale joint transit search on TESS 2-minute light curves, run on a 250k node-hour Blue Waters allocation. Read our successful proposal:

Blue Waters proposal

Other

Contact

Email: xiaziyna@gmail.com · tjamila@umich.edu