To this day I remember staring, almost skeptically, at a grainy black-and-white image of the HR 8799 system in a Science Olympiad test, and thinking: how can those scattered gray blobs be planets? As I dug into this intriguing question, my curiosity about other worlds and analyzing images turned into the scientific challenge of extracting exoplanet atmosphere features from starlight by combining statistical inference with physics-based models. It’s been 5 years since I wondered about this young main-sequence star and its four planets almost 133 light years away but this stellar mystery has hooked both the evidence-seeking and meaning-making parts of my brain. The evidence-seeking instinct pushed me towards NASA datasets, advanced spectroscopy papers, and reaching out to experts and mentors. I became fascinated with making reliable inferences which led me to pursue statistics and advanced math in school, and a mixture of self- and guided-studies in AI and data science. Based on my studies and outreach, I realized that this was an active area of research further fueled by the incredible observational capabilities of space telescopes like JWST. The joy that this exploration brought me and the collaborations I earned reaffirmed exoplanet atmosphere analysis as my area of interest. Yet, as I've pursued these interests, I've become increasingly aware of the barriers women continue to face in STEM. In my own classes, I've watched the number of girls dwindle as the science courses become more advanced, and I've counted how few of my math and physics teachers are women. At my internships at UC Irvine and NASA, mentors shared stories of their contributions being overlooked. These experiences reminded me that the challenges faced by women like one of my role models astronomer Henrietta Leavitt, whose groundbreaking work was not fully recognized during her lifetime, have not entirely disappeared. I don't want to simply navigate these barriers myself. I want to help dismantle them. Through programs with Girls Inc. Orange County, my peers and I are challenged stereotypes and pursuing leadership in science and engineering. At Caltech, I am a part of GWIPMA (Gender Minorities and Women in PMA) as well as the Society of Women Engineers where we hope to create community and open doors for women in STEM. I believe advocacy begins with making women's contributions visible, whether that means highlighting female scientists in classrooms, inviting them to speak, or encouraging girls to take on technical and leadership roles in research. I also want to advocate for institutions to examine who gets opportunities, whose voices are heard, and whose contributions are recognized. I have interviewed professors, written papers and posters, and even taught young kids to write short stories. These experiences have shown me that communication can do more than make science accessible. It can challenge assumptions, amplify underrepresented voices, and inspire people who might not otherwise see themselves as scientists. I want to use my skills as a communicator to speak up for women in STEM, share their achievements and experiences, and encourage the next generation of girls to pursue their scientific curiosity with confidence. Five years ago, a few gray blobs sparked my own journey into astrophysics. I hope to spend my career uncovering the mysteries of worlds beyond our own, while also helping build a scientific community where women have the opportunity not just to participate, but to lead.

I think the next major scientific advancement will be the ability to identify and interpret the atmospheres of potentially habitable exoplanets at scale using machine learning. Over the past few decades, astronomy has moved from asking whether planets exist beyond our solar system to confirming thousands of them. The next leap will be more difficult and more profound: determining what those planets are like, what molecules exist in their atmospheres, and whether any show conditions that could support life. My interest in this question began with a grainy black-and-white image of the HR 8799 system on a Science Olympiad test. Four faint dots circled a young star about 133 light-years away. I remember wondering how those gray blobs could be planets at all. That question eventually grew into a deeper one: how can scientists responsibly infer entire worlds from limited, noisy, indirect evidence? I explored this question by studying exoplanet atmospheric characterization. When a planet passes in front of its star, some starlight filters through the planet’s atmosphere. Different molecules absorb different wavelengths, leaving tiny patterns in the light. From those patterns, scientists can search for a plethora of compounds including water vapor, carbon dioxide, methane, and other molecules that may reveal a planet’s climate, chemistry, or habitability. The challenge is that these signals are extremely small and often buried beneath instrument noise, stellar activity, and complex atmospheric degeneracies. The future breakthrough will not come from telescopes alone, but from combining powerful observatories with machine learning, physics-based models, and careful statistical interpretation. During my NASA SEES internship and AGU Bright STaRS research, I explored this intersection by working on machine learning approaches for accelerating transmission spectroscopy. I studied how JWST data moves through the Eureka! pipeline, from raw detector files to light curves and transmission spectra. I also worked with synthetic spectra generated through atmospheric modeling tools and trained a dual-region convolutional neural network to identify molecules such as H2O, CO2, and CH4. What excited me was not just that a model could classify spectral features, but that it could help scientists search more efficiently through faint, complex data while still requiring careful validation. Globally, this advancement would affect society in a way that is both scientific and philosophical. The discovery of a strong biosignature candidate would not immediately answer whether we are alone, but it would change the scale of the question. It would require international collaboration, open data, public communication, and humility about uncertainty. Scientists would need to explain not only what was found, but how confident we should be, what alternative explanations remain, and what observations should come next. That is why I see this advancement as more than a technical milestone. AI-enabled exoplanet atmospheric characterization could help humanity move from planet detection to planetary understanding. It could teach us how common Earth-like conditions are, how atmospheres evolve, and whether life’s ingredients appear elsewhere in the galaxy. Most importantly, it would show how science progresses at its best: not by turning faint signals into easy certainty, but by turning them into better questions, better methods, and a more careful understanding of our place in the universe.