Emrul Hasan
Emrul Hasan
AI Researcher · UBC & BC Cancer
CTO / Technical Lead · LuxeFactor AI
Agentic AI Technical Specialist · Vector Institute

I am a Postdoctoral Research Fellow at the University of British Columbia (UBC) and BC Cancer, developing agentic AI systems for healthcare, including conversational assistants for cancer patients and clinicians. My research focuses on NLP model training and the design, evaluation, and safety of AI systems in high-stakes clinical settings.

As CTO and Technical Lead at LuxeFactor AI, I lead system architecture, technology selection, production validation, and secure AI adoption. I guide engineering and machine-learning teams, support hiring and knowledge transfer, and ensure evaluation and deterministic scoring systems follow approved methodologies.

At the Vector Institute, I have advised more than 15 companies on production agentic AI, including multi-hop RAG, tool-augmented agents, and reliability evaluation. Previously, as an Applied Scientist Intern at Amazon, I developed LLM-as-judge and LLM-as-jury pipelines that increased contact deflection by 2% and saved approximately four weeks of manual review.

My expertise spans NLP, agentic AI, and recommender systems, including model fine-tuning, retrieval, orchestration, and evaluation. I work with Python, PyTorch, Hugging Face, LangChain, and LangGraph, and deploy systems using AWS, GCP, Elasticsearch, and OpenSearch.

My research has appeared in ACM Computing Surveys, ACM RecSys, and IEEE venues.

Vector Institute
Advised 15+ companies
Amazon
Evaluated AI assistant
Research
Research Themes
NLP & GenAI
Research on natural language processing and generative AI — including large language models, RAG systems, text understanding, clinical NLP, and conversational AI for cancer care and mental health.
Agentic AI and Evaluation
Designing and evaluating LLM-based agentic architectures — multi-hop RAG pipelines, tool-augmented agents, LLM-as-Judge/Jury frameworks, and rigorous evaluation for production and clinical settings.
Recommendation Systems
Multi-criteria, review-based, and explainable recommendation using deep learning, contrastive learning, and NLP — with publications in ACM Computing Surveys (IF 30.4) and ACM RecSys.
Publications
Selected Publications
2025
Contrastive Learning for Aspect Representation towards Explainable Recommendation Best Student Paper
24th IEEE/WIC International Conference on Web Intelligence and Intelligent Agent Technology
2024
ACM Computing Surveys · Ranked 1/143 in CS Theory & Methods
2024
RecTour 2024 Workshop @ ACM RecSys · Bari, Italy
2024
arXiv preprint · Raza, Saleh, Hasan et al.
2022
IEEE International Conference on Big Data
View All Publications →