Research
I am interested in understanding how galaxies assemble and form their stars. Star formation is governed by the baryon cycle, operating on scales of tens to hundreds of parsecs: gas cools and condenses into molecular clouds, which form stars; young stars then inject energy and momentum into their surroundings through radiation, stellar winds, and supernovae, heating, dispersing, and enriching the gas; the transformed interstellar medium (ISM) then regulates subsequent star formation. These interconnected processes regulate the formation of new stars and shape the evolution of galaxies.
As an astronomer, I use ground- and space-based telescopes to observe different stages of the baryon cycle across diverse environments. Tracing these stages requires observations at different wavelengths, sensitive to different components of galaxies, including stellar populations, gas in its ionized, neutral, and molecular phases, and dust. The resulting datasets are complex and multimodal, making it challenging to fully exploit all the available information for physical inference and discovery. I develop machine-learning tools to integrate these complementary datasets and extract new physical insights.
A schematic sketch of the baryon cycle operating on 10-100 pc scales across galaxies. The galaxy in the middle is the beautiful IZw18, observed with the James Webb Space Telescope [Image credit: ESA/Webb, NASA, CSA, A. Hirschauer, M. Meixner et al.].
Below are some projects I am particularly excited about:
Mapping the ISM and Stellar populations of TinY galaxies (MISTY)
MISTY is a new multi-wavelength survey aimed at constraining the complex interplay between stellar populations and the multiphase ISM across the metal-poor regime. We are combining ground-based optical integral field spectroscopy (Magellan and Keck telescopes) with space-based Hubble and James Webb Space Telescope imaging to map star clusters, ionized gas (abundances and kinematics), and dust across a large, representative population of nearby dwarf galaxies.
Some of the MISTY galaxies, selected from the DESI dwarf galaxy catalog. The yellow, orange, and blue rectangles show the spatial coverage of the optical IFU we use.
Constraining the interstellar radiation field across massive discs (PHANGS)
The interstellar radiation field is a key agent in the baryon cycle, regulating the phase structure, chemistry, and thermal state of the ISM. I aim to constrain its intensity and spectral shape across diverse ISM environments, and to understand the stellar populations and physical processes that shape it.
As part of the PHANGS collaboration, I use its exceptional multiwavelength dataset of nearby massive galaxies, combining high-angular-resolution imaging and spectroscopy across the electromagnetic spectrum. These observations provide complementary tracers of the radiation field, including optical emission lines that probe the ionized gas and mid-infrared emission from polycyclic aromatic hydrocarbons (PAHs), which trace UV-irradiated regions shielded from ionizing radiation. By combining different tracers, I aim to map the radiation field across galaxies and understand what powers it and how it varies with the local environment.
A color composite of NGC 1566, using Hubble Space Telescope observations (top left corner) and James Webb Space Telescope (bottom right) by PHANGS. The full data traces stellar populations, multiphase gas (ionized, neutral, and molecular), and dust, including PAHs.
Building unified frameworks for astronomical discovery and inference with machine learning (FM-ISM)
Astronomy is entering an era of big data, with large-scale surveys mapping millions of astronomical objects across wavelengths and, sometimes, as a function of time. I develop statistical and machine-learning methods to mine these datasets and make new discoveries. In my view, one of the most exciting opportunities is to combine surveys observing the same objects through different wavelengths and data types (e.g., images, spectra) within a unified framework for inference and discovery.
Through FM-ISM, I am developing such a framework for nearby galaxies, among the most extensively observed objects in astronomy. We are building a probabilistic model of emission across wavelength and space, fitted jointly to multiwavelength images and spectral cubes of nearby galaxies. The model forward-models how astronomical emission is modified by distance and by the stochastic responses of the sky and individual instruments, enabling us to combine heterogeneous observations within a consistent physical framework.
This work is done in collaboration with the Center for Decoding the Universe (Stanford University).
Our goals include: (i) Uncover the rich correlation structure imprinted by the complex processes that connect stars, gas, and dust, through their multi-wavelength observables. (ii) Enable inference using all the data without downgrading the high-resolution data to the worst resolution. (iii) Enable data-driven science while accounting for noise, upper limits, varying angular resolution, and depth.
Recorded Talks
Below you can find some of my recorded talks.
From data to insight: machine-assisted discoveries in the big data era in astronomy
Steward Observatory-NOIRLab joint colloquium & TAP Computation Lecture
March, 2026.
Link to watch on panopto.
Encoding the nearby Universe with FM-ISM
Cosmic-AI seminar @ NOIRLab.
March, 2026.
The Role of AGN-Driven Winds in Their Host Galaxies' Evolution
Caltech Colloquium
February, 2023.
Extracting the Main Trend in a Data Set: The Sequencer Algorithm.
AAS Journal Author Series @ YouTube.
November, 2021.
Finding simple structures in complex astronomical datasets.
Astroinformatics 2018.
Heidelberg.