I am a postdoctoral researcher at the University of Central Florida (SST Lab). My work focuses on AI-driven transportation safety, spanning connected-vehicle (CV) trajectory & video analytics, causal and behavior-aware safety modeling, and spatiotemporal heterogeneity for proactive crash prediction and risk assessment.

Selected Journal Publications (by Theme)

1) Causal & Explainable AI for Safety

• Transportation Research Part C (2025): ML + causal mediation for interpretable crash risk mechanisms using CV data.
  TR-C, 183, 105479. https://doi.org/10.1016/j.trc.2025.105479
• Transportation Research Part C (under review / in progress): Behavior-aware crash prediction with macro–micro exposure fusion (traffic + CV pre-crash dynamics).
  (Manuscript line: dual-level exposure + causal heterogeneity)

2) Spatiotemporal Heterogeneity & Bayesian Safety Modeling

• Analytic Methods in Accident Research (2025): Grouped random-parameters Poisson–Lindley with spatial effects for intersection crash analysis.
  AMAR, 47, 100387. https://doi.org/10.1016/j.amar.2025.100387

3) Crash Type & Injury Severity Joint/Multivariate Modeling

• Analytic Methods in Accident Research (2025): Joint analysis on pedestrian injury severity across vehicle movements at intersections: Addressing temporal instability and spatial correlations. AMAR, 47, 100406. https://doi.org/10.1016/j.amar.2025.100406

4) ITS Operations, Tunnel Safety & Multi-task Learning

• Accident Analysis & Prevention (2025): Cross-stitch networks to jointly model tunnel crash severity and congestion duration (safety–mobility coupling).
  AAP, 213, 107942. https://doi.org/10.1016/j.aap.2025.107942

5) Connected & Automated Vehicles (CAV) and Cybersecurity-Oriented Safety

• Review / survey line (selected): Cyberattack patterns, detection, stability impacts, and robust control strategies for CAV fleets in complex traffic scenarios.

6) Multimodal / Video Understanding & Risk (VLM / LLM / Diffusion)

• Scientific Data (under review): SAVeD dataset for ADAS-equipped near-miss & crash events to enable multimodal understanding and counterfactual safety analysis.
• Methods line (in progress): LLM-assisted event parsing + controllable diffusion world models for counterfactual “what-if” safety assessment.

Collaboration

I’m always open to collaboration on CV trajectory analytics, Bayesian/causal safety modeling, and multimodal risk understanding.