Zaer Bin Zaqy

Academic Interests

Math & Computer Science

Fun Fact

I have the same birthday as my younger brother!

A Malay boy from the little red dot of Singapore, I’ve rarely heard myself on television—epiphany struck hit in waves, 1.2 terabytes of them. While building a Singlish (Singapore-English) text-to-speech app for the local deaf community at the Singapore Engineering Good Tech-ForGood Festival, I leaned on various speech corpora, data sets of local speech. Sifting through each ‘.wav’ file, reality hit: I could barely hear myself. Among the terabytes of Singaporean voices, the distinct cadence of my Malay heritage was eerily absent. More than a data gap, this was cultural silence.

As a Malay speaker, Bahasa Melayu, my native language, is considered a “low-resource” language, with far fewer tools and datasets than English or Mandarin. Out of 7,000+ global languages, only about 20 have robust computational resources, a disparity that mirrors the underrepresentation of ethnic minorities like mine in media and technology. That absence in digital spaces isn’t just a technical oversight; it reinforces whose voices get amplified and whose fade into the background.

At Duke, where I study Mathematics and Computer Science with a growing interest in philosophy, I’ve found myself drawn to the theoretical side of computing, complexity theory, formal language theory, automata. These fields captivate me not only for their elegance but also for how they reveal the boundaries of what can be represented and computed. To me, that’s not abstract, it’s deeply connected to questions of linguistic representation, accessibility, and cultural identity.

Where I began with NLP and representation, I’ve since found myself increasingly interested in interpretable machine learning, low-latency ML infrastructure, causal inference, and Bayesian networks. These directions excite me because they combine rigor with impact: building systems that are both technically sound and human-centered.

I want my research and technical work to carry those questions forward, how computation can better serve diverse voices, and how lived experiences can shape the way we design systems. Beyond Duke, I hope to continue this journey in graduate study and research, contributing to a digital world where no ‘.wav’—however marginalized—goes unheard.

What do you like about being a SPIRE Fellow?

SPIRE offers the kind of academic home and community I’m looking for—a place where I can find mentorship and support while contributing to a community that celebrates diversity in STEM. This morning, I read a paper on scaling human-centered machine translation, leverages sparsely gated mixture-of-experts models to optimize multilingual neural machine translation across 200 language pairs. The paper was coauthored by over 264 authors from 3 different institutions. That’s what STEM looks like in 2024: communal and collaborative.

I am a committed researcher. I believe the most fascinating discoveries often happen in the interstices of fields. Quantum computing, for example, slides seamlessly between computer science and physics, while formal language theory sits at the intersection of computer science and linguistics. Just this past week, I had the privilege of geeking out over a quantum natural language processing paper I’d read with one of my talented SPIRE Fellow mates, a physics genius from Kenya, with no prior physics knowledge on my end whatsoever. This spontaneous exchange epitomizes the kind of vibrant, interdisciplinary and diverse environment that I love that comes with being a SPIRE Fellow.