1<!DOCTYPE html> 2<html lang="en"> 3<head> 4 <meta charset="utf-8"/> 5 <meta name="viewport" content="width=device-width, initial-scale=1"/> 6 <title>About Me - Lucia Zheng</title> 7 <link rel="stylesheet" href="https://edwardtufte.github.io/tufte-css/tufte.css"/> 8 <link rel="stylesheet" href="main.css"/> 9</head> 10<body> 11 <article> 12 <h1>About Me</h1> 13 14 <nav> 15 <ul> 16 <li><a href="index.html">â Home</a></li> 17 </ul> 18 </nav> 19 20 <section> 21 <p>Hi, I'm Lucia. I'm a JD/CS PhD student at Stanford, advised by <a href="https://law.stanford.edu/directory/daniel-e-ho/" target="_blank" class="nowrap">Daniel E. Ho</a>, 22 <a href="https://nlp.stanford.edu/~manning/" target="_blank" class="nowrap">Christopher Manning</a>, 23 and <a href="https://www.peterhenderson.co/" target="_blank" class="nowrap">Peter Henderson</a>, at the <a href="https://reglab.stanford.edu/" target="_blank">RegLab</a>, <a href="https://nlp.stanford.edu/" target="_blank">Stanford NLP Group</a>, 24 <a href="https://www.polarislab.org/" target="_blank">Polaris Lab</a>, and <a href="https://citp.princeton.edu/" target="_blank">CITP</a>. I'm supported by the <a href="https://vpge.stanford.edu/fellowships-funding/sigf" target="_blank">Stanford Interdisciplinary Graduate Fellowship</a>. 25 </p> 26 27 <p>My research focuses on the intersection of maching learning, NLP, and law. 28 I'm interested in improving language modeling for reasoning in environments 29 where information is incomplete and shaped by competing interpretations, norms, and institutions,
30 with an eye towards supporting public interest applications. 31 </p> 32 33 <p> 34 My past work has centered on legal reasoning, where I studied how legal contexts 35 expose core challenges in language model capabilities, including analogical reasoning, 36 interpretive ambiguity, epistemic calibration, and collaborative decision-making. 37 </p> 38 39 <p> 40 More recently, I've been working on methods for principled information-seeking in 41 language model agents and open-ended discovery problems. And more broadly, thinking about 42 belief representations, connections between reinforcement learning theory and language modeling, 43 and how to develop better computational models of evidence-driven learning that support hypothesis 44 formation, critique, and revision, inspired by ideas from philosophy and cognitive science. 45 </p> 46 47 <p> 48 I'm also interested in technical and regulatory questions around AI governance, especially questions of traceability and quantifying the contribution of platform and provider-side design choices to downstream effects. 49 </p> 50 51 <p> 52 In law school, I lead the <a href="https://law.stanford.edu/stanford-animal-protection-project-sapp/" target="_blank">Stanford Animal Protection Pro Bono Project</a> (SAPP), 53 serve as president of the <a href="https://law.stanford.edu/slata/" target="_blank">Stanford Law and Technology Association</a> (SLATA), 54 as VP of the <a href="https://law.stanford.edu/stanford-artificial-intelligence-law-society-sails/" target="_blank">Stanford Artificial Intelligence and Law Society</a> (SAILS), 55 as an Academic Co-Chair of the <a href="https://www.acslaw.org/" target="_blank">American Constitution Society</a> (ACS), 56 and am a member editor on the <a href="https://www.stanfordlawreview.org/" target="_blank">Stanford Law Review</a>. 57 </p> 58 </section> 59 60 <details> 61 <summary><h2>Other Past Lives</h2></summary> 62 <p> 63 I've built computational tools for administrative adjudication, legal and policy analysis, and regulatory enforcement, 64 through collaborations with the <a href="https://www.dol.gov/" target="_blank">U.S. Department of Labor</a>, the <a href="https://www.nj.gov/defender/" target="_blank">New Jersey Office of the Public Defender</a>, and the <a href="https://www.aclu.org/" target="_blank">ACLU</a>. 65 In law school, I've done research on air monitoring and air permit compliance in the San Joaquin Valley for the <a href="https://leadershipcounsel.org/" target="_blank">Leadership Counsel for Justice and Accountability</a><label for="sn-sjv" class="margin-toggle sidenote-number"></label><input type="checkbox" id="sn-sjv" class="margin-toggle"/><span class="sidenote">See our <a href="https://drive.google.com/file/d/1C0Zs8rlwZN4FarnnzKViLs4KKrfIoW45/view?usp=sharing" target="_blank">legal brief</a>, <a href="https://drive.google.com/file/d/1-FgMsB6uMScR91r7PHoPh3XNUInBvELA/view?usp=sharing" target="_blank">technical plan</a>, and <a href="https://dreamy-salamander-39eb0e.netlify.app/map.html" target="_blank">interactive map</a>.</span> and language model-based methods for identifying fact patterns in environmental litigation for <a href="https://www.animalpartisan.org/" target="_blank">Animal Partisan</a>. 66 </p> 67 68 <p> 69 Prior to graduate school, I worked on machine learning for the <a href="https://www.linkedin.com/" target="_blank">LinkedIn</a> newsfeed, 70 on backend infrastructure at <a href="https://www.khanacademy.org/" target="_blank">Khan Academy</a> during the COVID-19 transition to remote learning, 71 on content integrity tools at <a href="https://about.meta.com/" target="_blank">Meta</a>, and as the first software engineer intern at <a href="https://changeresearch.com/" target="_blank">Change Research</a>, a public benefit corporation startup, on making representative online polling more accessible to grassroots and downballot political campaigns using statistical techniques, where I contributed to a <a href="https://patents.justia.com/assignee/change-research-public-benefit-corporation" target="_blank">patent application</a> on my work<label for="sn-gillum" class="margin-toggle sidenote-number"></label><input type="checkbox" id="sn-gillum" class="margin-toggle"/><span class="sidenote">We were the first polling firm to correctly call Andrew Gillum's win in the Florida Democratic primary.</span>. 72 </p> 73 74 <p> 75 I graduated with a B.S. in Computer Science and a minor in Math, and an M.S. in Computer Science 76 from Stanford, advised by <a href="https://sites.google.com/site/marywootters/" target="_blank">Mary Wootters</a>. 77 </p> 78 79 <p> 80 I spent about six years as a volunteer teacher and mentor at <a href="https://streetcode.org/" target="_blank">StreetCode Academy</a>, 81 a non-profit organization that provides free computer science and entrepreneurship education in East Palo Alto. 82 </p> 83 84 <p>
85 As a high school student, I was the Pacific Northwest Director of Activism for Junior State of America, where I registered thousands of high school students to vote through <a href="https://nationalvoterregistrationday.org/" target="_blank">National Voter Registration Day</a>. 86 I was also a field organizer for The Bus Project (now Next Up Action Fund), a non-profit organization focused on getting young people engaged in politics and voting issues. 87 I testified before the Secretary of State of Oregon Jeanne Atkins on behalf of The Bus Project in support of the <a href="https://sos.oregon.gov/voting/Pages/motor-voter-faq.aspx" target="_blank">Oregon Motor Voter Act</a>, passed in 2016. 88 </p> 89 </details> 90 91 <section> 92 <p> 93 <a href="connect.html">Connect</a> · 94 <a href="reading.html">Reading</a> 95 </p> 96 </section> 97 </article> 98</body> 99</html> 100
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.