AI Scientist
My work spans language, vision, retrieval, privacy, and healthcare, with a focus on systems that must handle ambiguity, learn from feedback, work across modalities, operate under constraints, and remain useful at scale.
Themes
I work on AI systems that improve from interaction and feedback without becoming brittle. That includes memory, personalization, retrieval, contextual adaptation, and the design of learning loops that remain reliable over time.
I am interested in models that connect language, vision, and structured evidence to interpret messy real-world inputs. This includes multimodal understanding, extraction, contextual perception, and visual reasoning about what matters in a scene.
Some AI systems operate in settings where privacy exposure, weak oversight, or poor grounding are unacceptable. I work on systems for safety, privacy, traceability, and human review in high-stakes workflows.
I care about models that survive deployment at scale. That means large-scale inference, robustness under operational constraints, and system design that keeps modeling ideas useful in production rather than only on benchmarks.
Selected publications
Mobile Cloud Visual Media Computing / 2015
CVPR / 2015
RepL4NLP at ACL 2022 / 2022
Service
Reviewer and judge for AI, NLP, statistics, and innovation venues, including NeurIPS, AISTATS, COLING, LREC, CES Innovation Awards, and QS Reimagine Education.
Contact
I am open to research conversations, collaboration, speaking, and advising around AI systems that need to work under real-world constraints.
