Publications &
Patents
Currently focused on knowledge graph-augmented retrieval for long-context financial documents, Graph Transformers, and cost-efficient multimodal LLMs for mixed-modality document understanding.
// patents
Method for Cyber Threat Detection Using News Alerting Language Models
Patent covering the CANAL framework — NLP-based cyber activity monitoring and alerting system that identifies emerging cybersecurity threats from news sources without relying on expensive LLM APIs.
System and Method for AI-Driven Financial News Alerting and Sentiment Analysis
Patent covering the FANAL framework — automated detection and alerting of financially significant news events using language models, applied to real-time market intelligence.
Multimodal Retrieval-Augmented Generation for Financial Document Question Answering
Patent covering the MultiFinRAG framework — an optimized multimodal RAG architecture for financial QA across mixed-modality documents including tables, charts, and text.
// publications