Huacen Xu |

About Me

Hi, I’m a Data Scientist and Data Engineer at BlackRock in Atlanta, where I work on Aladdin Data — the data platform behind Aladdin, BlackRock’s investment and risk platform. I build the engineering layer of data governance: the compliance pipeline that turns policy SQL into per-dataset compliance status, the certification component of Certification-as-a-Service (a Streamlit platform generating PDF certificates), governance agent skills grounded in the internal policy wiki, and the target-state architecture for the enterprise data taxonomy. I lead the firmwide data taxonomy approval forum, and I’m a co-inventor on a patent-pending tokenized data-distribution framework.

Before that I was a research analyst at the University of Georgia, where I completed an MS in Business Analytics, building NLP pipelines over 85K+ news articles and modeling social and economic questions with multilevel models and decomposition methods.

Focus Areas

Governance Engineering & Automation: Building the tooling behind enterprise data governance — compliance pipelines, certification automation, taxonomy change management, and AI-grounded agent skills.
Agent Evaluation & Applied ML: LLM agents, retrieval-augmented generation, embedding models and vector search; token-usage and prompt-caching evaluation harnesses; NLP and predictive modeling applied to social and economic questions — peer-reviewed research on moral framing, credit access, and housing disparities.
Data Tokenization (US Patent Pending): Co-inventor on a blockchain-based approach to tokenizing data access entitlements for enterprise data governance, where the token balance is the access control and the ledger is the audit trail.

Governance Engineering Patterns

Worked out on sanitized synthetic data, as one architecture:

Compliance-check pipeline — Governance policy checks as SQL files, run through a fixed extract → transform → metrics → load pipeline; a new domain is config only, no core code changes.
Certification workflow — Turns the pipeline’s output into a certificate PDF, walks it through a two-stage approval with HMAC-signed one-click links, and grounds a Q&A assistant on a resilient LLM client (model fallback, error triage, circuit breaker).
Governance knowledge graph — Syncs a policy wiki into a deterministic knowledge graph where every edge explains itself, routes questions by walking the graph, and gates the knowledge base with a three-score eval in CI; exposed as agent skills for answering policy questions and evaluating taxonomy changes.
Architecture notes — Where each piece sits on an enterprise data platform, how they connect through the catalog, and what’s next.

Pattern code available on request.

Selected Publications

Student Loan Debt and the Black-White Homeownership Gap — Huacen Xu, C.J. Dawkins, M. Travis P., M. Chowdhury - Housing & Society (under review), 2026. Predictive modeling of how student loan debt shapes racial disparities in homeownership — quantitative methods applied to a structural economic question. Code.
Convergence or Divergence? A Cross-Platform Analysis of Climate Change Visual Content Categories, Features, and Social Media Engagement on Twitter and Instagram — S. Qian, Y. Lu, Y. Peng, C.C. Shen, Huacen Xu - Public Relations Review, 2024. Unsupervised image clustering identified five categories of climate visuals and platform-specific engagement patterns — ML-driven visual content analysis at scale.
Examining Public Attitudes and Ideological Divides Through Media Engagement: An Empirical Analysis of Moral Foundations Theory Amidst the COVID-19 Pandemic — Huacen Xu - Current Research in Social Psychology, 33(2), 2024. Moral-foundations scoring of partisan and science media posts, including a new liberty dictionary, to model how moral framing predicts audience engagement. Code.
Can Microfinance Achieve Both Social and Financial Goals? Investigation on MicroCred Nanchong — Huacen Xu - Inclusive Finance in China (Palgrave Macmillan), 2021. Field investigation of how microfinance institutions balance social and commercial objectives.
Double Bind in Loan Access in China: The Reification of Gender Differences in Business Loans — Huacen Xu, H.J. Zhan, C.E.E. James, L.D. Fannin, Y. Yin - International Journal of Gender and Entrepreneurship, 2018. Predictive modeling of gender disparities in loan access — how perceived rejection likelihood suppresses applications.

Full publication list on Google Scholar. Code and projects on GitHub.