Firefighting remains one of the most hazardous professions in the United States. Firefighter fatalities and injuries, particularly sudden cardiac death (SCD) and those caused by sudden fire behavior, remain unacceptably high. Today, firefighters largely rely on experience, training, and intuition to make time-critical decisions under extreme uncertainty, with limited real-time analytical support. Existing physics-based and data-driven prediction models are often impractical for operational use due to unrealistic assumptions, limited data availability, and prohibitive computational costs.

To address these challenges, the Fire Research Division of NIST’s Engineering Laboratory is developing the next generation of AI-enabled firefighting decision-support systems. Our goal is to deliver real-time, computationally efficient, trustworthy, and interpretable intelligence that can be used in 1) firefighter health monitoring, 2) during pre-emergency planning, and 3) active incident response.

We are seeking individuals to develop novel machine-learning frameworks that integrate multiple AI paradigms to enhance situational awareness, operational effectiveness, and firefighter safety in three crucial research areas:

1. Interpretable Deep Learning for Firefighter Health and Sudden Cardiac Death Prevention

• Develop advanced, interpretable deep-learning models using wearable sensor data (e.g., ECG, heart rate, activity, thermal exposure) to:
  • Detect early physiological warning signs associated with sudden cardiac death and overexertion.
  • Provide real-time or near-real-time alerts that firefighters and incident commanders can trust.
• Prioritize model interpretability, robustness, and validation, ensuring insights can be understood and acted upon in high-stress operational environments.

2. Large Language Models for Incident Reporting and Pre-Emergency Intelligence

• Develop and adapt LLMs to:
  • Automatically synthesize incident reports, sensor streams, and historical fire data into actionable summaries.
  • Support pre-incident planning, risk assessment, and knowledge retrieval for fire departments.
  • Enable natural-language interaction between firefighters, incident commanders, and AI systems during rapidly evolving emergencies.
  • Address challenges in hallucination mitigation, uncertainty quantification, and domain-specific grounding for safety-critical applications.

3. Recommendation Systems for Fireground Strategy and Search & Rescue

• Design AI-driven recommendation systems that provide:
  • Optimal and/or safest strategies for firefighting, ventilation, and search-and-rescue operations.
  • Dynamic decision support that adapts to changing fire conditions, building layouts, and resource constraints.
• Incorporate multi-objective optimization balancing fire suppression effectiveness, time-to-rescue, and firefighter safety.
• Emphasize human-in-the-loop decision-making, ensuring recommendations are transparent, explainable, and actionable.

[1] Li, J., Brown, C., Dzikowicz, D.J., Carey, M.G., Tam, W.C. and Huang, M.X., 2023. Towards real-time heart health monitoring in firefighting using convolutional neural networks. Fire Safety Journal, 140, p.103852.

[2] Agarwal, N., Shrivastav, S., Henneman, J., Knapp, C., Cushman, J., Aktas, M., Xia, X., Hostler, D., Carey, M.G., Tam, W.C. and Dzikowicz, D.J., 2025. Performance of a wearable device for real-time cardiac monitoring among active firefighters. Journal of Electrocardiology, 91, p.153961.

[3] Fang, H., Zhang, B., Tam, W.C., Yang, C. and Lo, S.M., 2025. A deep learning-based approach for unsafe area prediction in building fire evacuation. Journal of Building Engineering, p.113723.

[4] Tam, W.C., Fu, E.Y., Li, J., Peacock, R., Reneke, P., Ngai, G., Leong, H.V., Cleary, T. and Huang, M.X., 2023. Real-time flashover prediction model for multi-compartment building structures using attention based recurrent neural networks. Expert Systems with Applications, 223, p.119899.

Large Language Models (LLMs); Recommendation Systems; Interpretable Deep Learning; Real-Time Decision Support; Wearable Sensors; Fireground Intelligence; Human-AI Teaming

citizenship

Open to U.S. citizens

level

Open to Postdoctoral applicants