Debanjan Ghosh

Hello All. I am a Principal Research Scientist at Analog Devices (ADI) Edge AI team. At ADI, my team is focused on developing LLM-based Agentic solutions - including deep research, coding tools, and small models - for internal engineers and customers supported by rigorous benchmarking experiments. We are also involved in building multimodal reasoning models for the next generation Edge.

Before, I was at Educational Testing Services (ETS), I led a team dedicated to Generative AI product development, creating content such as passages, social conversation, Q&A for learning apps, and traditional assessments. I also contributed to building models for automated scoring of GRE/TOEFL essays.

Previously, I worked as a Senior Software Engineer at Thomson Reuters Research in NYC, where I was involved in a range of information retrieval and extraction projects for the Westlaw legal search engine and Reuters news, including classification and recommendation systems.

Prior to returning to industry, I did a Ph.D. investigating contextual knowledge—particularly figurative language like sarcasm and irony—on social media platforms. I also completed a postdoctoral fellowship at MIT’s Brain and Cognitive Sciences Department, studying how semantic knowledge is represented in the brain.

I am a co-organizer of the SemEval Workshop for 2025-2026, which is among the largest workshops in NLP in terms of tasks and number of participants. Be sure to explore the tasks and proposals for the new challenges that focus on evaluating systems for computational semantic analysis.