Summary
Adib Mosharrof is a PhD candidate and Graduate Research Assistant at the University of Kentucky with 11 years of experience building multimodal and task-oriented dialogue systems. He develops proactive conversational agents that integrate Vision-Language Models like SmolVLM2 to understand procedural videos, track step-wise progress, and summarize instructional tasks through the ProAssist project. His work on end-to-end zero-shot task-oriented dialogue and schema augmentation is grounded in interpretability-driven neural analysis to improve generalization without turn-level annotations. Previously he improved factual consistency for summarization at Prime Video / Amazon MGM by designing fact decomposition and Multi-Instance Learning verifiers, showing a practical bridge between research and production-scale ML. He also brings a strong applied vision background from DARPA-funded media forensics and video summarization projects, and has publications in WWW, SIGIR, FLAIRS and EMNLP workshops. Based in Lexington, KY, he combines deep academic rigor with hands-on engineering across large-scale training pipelines and model interpretability.
11 years of coding experience
6 years of employment as a software developer
Master's degree, Computer Science, Master's degree, Computer Science at University of Oregon
Bachelor of Science (B.Sc.), Computer Science, Vice Chancellor's Gold Medal, Bachelor of Science (B.Sc.), Computer Science, Vice Chancellor's Gold Medal at BRAC University
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of Kentucky
English, Bengali