Data & Intelligence Lab

Research

Research Directions

• AI & Information Retrieval

  We study LLMs as retrieval agents for dataset discovery and evidence finding, combining metadata augmentation, hybrid search, GraphRAG-style provenance, and agentic search when resources are sparse or weakly described.

• LLM-in-the-middle for Physical AI

  We explore how multimodal foundation models can sit between perception, planning, and control in safety-critical settings, with an emphasis on semantic reasoning, traceability, and runtime safety supervision.

• AI interpretability

  We work on interpretable machine learning for systems that need to be understood, audited, and trusted, including feature attribution, diagnostics, and explanations that stay useful as data shifts over time.

• AI in Education

  We study how AI can support secure, scalable, and meaningful digital learning environments. Our work addresses hands-on assessment, digital skills, and the challenges generative AI creates for traditional exams, while keeping governance, ethics, and responsible data use central.

• Multimodal AI and
  environmental monitoring

  This area brings together Earth observation, sensor fusion, and digital-twin analysis. We study image segmentation, anomaly detection, and AI-supported monitoring of changing environments.

• NLP and semantic
  password forensics

  We study NLP methods for password analysis and related security problems. Word embeddings, semantic retrieval, and retrieval-augmented methods reveal structure in user-generated secrets.

• Digital health and
  wearable sensing

  We work with mobile and wearable data in rehabilitation and stress monitoring, with a focus on adaptive models that can learn from noisy real-world signals and patient-generated data.