About me

I’m a professor at NYU’s Tandon School of Engineering, with appointments in Computer Science and Data Science, and I co-direct the VIDA Center. My research focuses on making data more accessible and trustworthy — spanning data analysis and integration, visualization, AI/ML, and information discovery, with applications in urban analytics, computational reproducibility, biomedical research, and online crime.

News

• I am honored to have received the VLDB Women in Database Research Award. I gave an invited talk and contributed an invited paper, “Semantic Data Systems: From Data Management to Data Understanding” (PVLDB 19(12), pp. 4973–4980).
• VIDA had a strong presence at VLDB 2026:
  • Grace Fan, Eden Wu, and Majid Daliri had a paper in the main conference, “MosaicJoin: Compact Semantic Sketches for Value-Level Join Discovery” (PVLDB 19(11), pp. 3745–3758)
  • Christos Koutras and collaborators presented a tutorial on biomedical data harmonization, “Interoperability in Healthcare: A Primer and New Frontiers” (PVLDB 19(12), pp. 4912–4917)
  • I gave two workshop keynotes:
    • “Semantic Data Systems: From Data Understanding to Agentic Workflows” (DashSys) and
    • “Hammers for Biomedical Nails: From Matching to Meaning in Data Harmonization” (BioDMS)
  • We presented three more papers at VLDB workshops:
    • “AgentTrails: Towards Trust and Reuse for Agentic Tasks” (DashSys),
    • “Quantizing Sampling-Based Inner Product Sketches” (VecDB), and
    • “Data Gatherer Revisited: Scalable Dataset Reference Extraction from Biomedical Literature” (BioDMS).
• Our NSF-funded project that aims to combat wildlife trafficking in online marketplaces was featured in Forbes.
• At SIGMOD 2026, we presented demos of two tools we have developed to support data harmonization:
  • BDI-Kit Demo: A Toolkit for Programmable and Conversational Data Harmonization — GitHub · demo video · docs
  • BDIViz in Action: Interactive Curation and Benchmarking for Schema Matching Methods — GitHub · live demo
• Our paper describing AutoDDG has been accepted to ACM SIGMOD 2026. AutoDDG combines traditional profiling and LLMs to automatically generate textual descriptions of structured data. The descriptions not only improve dataset findability, but also make it possible for LLMs to understand datasets. Here’s the preprint: https://arxiv.org/pdf/2502.01050. The system is open source and available at https://github.com/VIDA-NYU/AutoDDG