Avant-garde & experimental.

Margarita Geleta

[linkedin] [@3omni] [x]

Geleta is an artist and AI/HCI researcher based in San Francisco. With a multifaceted research background, spanning audio + spatial computing and AI/ML for population genetics, Geleta led research projects at the Berkeley AI Research laboratory and the Stanford Department of Biomedical Data Science. Geleta co-developed AudioMiXR, an XR interface for spatial audio manipulation, and authored PixInWav, an image-in-audio steganographic method for covert communication. In 2025, she is joining Microsoft Research audio & acoustics research group. Previously, Geleta interned as a sound tech researcher at Dolby Laboratories at the Advanced Technology Group, and as an applied scientist at Amazon's Home Innovation Team, specializing in large-scale generative models and GAN inversion. Geleta also co-founded a defense tech startup backed by Founders, Inc.

Did I say I'm finishing my CS PhD at UC Berkeley?

Recent projects [DM for collab]

Audio mixing in augmented reality on Apple Vision Pro. Demo recorded on stage at Temple Nightclub in San Francisco. [Read more]

PCB/GPU fashion statements. Designer purses using PC hardware components and jellyfish pigments. Contact me for more details. [Gallery]

Events hosting & speaking engagements. Too social of a person, so much that I've co-organized dozens of tech workshops, hackathons, career fairs, and conferences. I am also available for speaking to tech audiences. [List]

I'm a 3rd generation audiophile. Everything started in the 80s. The tape markets, the Radio Engineering magazines, and the love for music. [Genealogy]

AudioMiXR [link to preprint]

We present AudioMiXR, an augmented reality interface intended to assess how users manipulate virtual audio objects situated in their physical space using six degrees of freedom (6DoF) deployed on a head-mounted display (Apple Vision Pro) for 3D sound design. We conducted an exploratory study where we recruited 27 participants, consisting of expert and non-expert sound designers. The goal was to assess design lessons that can be used to inform future research venues in 3D sound design. [Link to project website]

Ancestry research [gscholar]



Genealogy has been my life-long passion, motivating me to pursue a PhD and reconnect with my extended family. My PhD thesis focuses on ancestry inference and investigative genetic genealogy, while my master’s thesis explored genotype simulation conditioned on pedigrees. Currently, I am interested in applying AI/ML techniques to enhance kinship prediction. If you think we are distant cousins, send me an e-mail.

Inferring continuous population structure coordiantes along the genome. A new paradigm for ancestry inference. [Under review]

Autoencoders for genetic variation. Autoencoders for genotype compression, simulation, and global ancestry inference. [Intro here, still under review]

Generative moment matching networks for genotype simulation. SNP simulation & privacy-preserving data sharing for biobanks. [Read here]

Pedigree-aware genotype simulation. A forward-in-time genotype simulator with integrated genealogical structure & sex-specific recombination. [Not public]