My updates on research, talks, product work, and projects.

Oracle published Safety One Argus 2026.1.01 release notes, including updates around AI-assisted adverse-event intake. The release connects closely to the pharmacovigilance AI work I have been doing at Oracle.

I gave a talk for IEEE Seattle on Reinventing Reverse Logistics: How AI is Turning Niche Processes into Scalable Circular Systems.

I presented My DRAGON paper on graph- and ontology-grounded agents for drug-safety review at the AAAI Workshop on Health Intelligence in Singapore as part of the W3PHIAI-26 program.

I presented AI for Drug Safety Case Review Using Graphs & Ontologies at Health Tech Week in San Francisco.

I spoke at IEEE New Era AI 2025 in Seattle in the retail track on Reinventing Reverse Logistics: How AI is Turning Niche Processes into Scalable Circular Systems, and won the Best Speaker Award.

Our DRAGON paper on graph- and ontology-grounded agents for drug-safety review was accepted to the AAAI Workshop on Health Intelligence.

Oracle’s VP of Drug Safety presented my Medical Safety Assessment work at World Drug Safety Congress events in Amsterdam and Boston.

I presented Pharmacovigilance AI for Medical Safety Assessment virtually at Oracle Health and Life Sciences Safety Summit EMEA, held in Utrecht, Netherlands.

Protopia AI received a $1.25M Direct-to-Phase II U.S. Air Force contract to demonstrate Stained Glass Transform for secure use of proprietary code with coding LLMs. My work there contributed to that broader privacy-preserving AI effort.

Protopia AI took first place in the U.S. Army xTechScalable AI 2 competition for privacy-enhancing AI work built around Stained Glass Transform. My work there contributed to that broader effort.

I contributed to Protopia AI’s AWS deployment work around Stained Glass Transform for enterprise LLM acceleration and privacy-preserving LLM use.

Our paper PersonalTM: Transformer Memory for Personalized Retrieval was accepted to SIGIR 2023.

Our paper Self-Aware Feedback-Based Self-Learning in Large-Scale Conversational AI was accepted to NAACL 2022.

Our paper A Vocabulary-Free Multilingual Neural Tokenizer for End-to-End Task Learning was accepted to RepL4NLP at ACL 2022.

Our paper Personalized Query Rewriting in Conversational AI Agents was released on arXiv.

The Boston Globe covered VIP: Finding Important People in Images, our computer-vision work on identifying socially important people in photographs using visual and contextual cues.

CloudCV was accepted into Google Summer of Code on March 2, 2015, and I joined the effort as a mentor.

Our paper VIP: Finding Important People in Images was accepted to CVPR 2015.

Our CloudCV chapter, Large-Scale Distributed Computer Vision as a Cloud Service, was published in Mobile Cloud Visual Media Computing.

We demoed CloudCV live at CVPR as a way to make computer-vision algorithms easier to access, run, and share through the cloud.