My research philosophy is driven by a commitment to enhancing community resilience and a deep interest in the complex dynamics of disaster response. I believe in a holistic, data-driven approach that considers the intricate interrelationships within a community system. For me, research is not just an academic exercise but a tool for real-world impact, which is fundamental to my work at the OU Community Resilience Research Laboratory, a part of the NIST Center of Excellence at Colorado State University.
I am particularly interested in developing innovative optimization models to minimize community loss after natural disasters. By analyzing data within these models, we can identify key factors that contribute to a community's resilience. My research also focuses on addressing the decisions of multiple stakeholders, leading me to create efficient algorithms for bi-level optimization problems.
My research is an iterative process, guided by theoretical concepts but grounded in empirical observations and validation. I believe that collaboration, interdisciplinary insights, and a commitment to constant learning are essential for generating meaningful, transformative outcomes. In essence, my philosophy is based on the idea that through rigorous research, we can empower communities and significantly improve their ability to withstand and recover from disasters.
My group studies how to make communities and infrastructure more resilient to natural hazards using optimization and machine learning with clear engineering assumptions. The goal is simple: turn hazard and asset data into transparent decisions for mitigation and recovery that balance cost, performance, and equity.
What we work on
Physics-informed impact modeling. We use standard fragility and vulnerability functions (earthquake, tsunami, flood) to translate hazard intensity and asset attributes into damage and functionality loss. This creates interpretable features instead of black-box labels.
Prediction with structure. We build representation-learning models (including graph models when networks matter) to estimate outcomes such as downtime, expected loss, and number of impacted people. We report calibration (e.g., ECE/Brier) and uncertainty bands, not just accuracy.
Decision optimization. We solve resource-allocation problems for retrofits and recovery. Typical objectives: minimize expected loss and reduce inequity using per-building normalized indices (e.g., Theil). We generate Pareto frontiers so stakeholders can see trade-offs.
Reproducible tools. Clean data dictionaries, config-driven code (Python/Gurobi), and figure scripts. Parts of our workflow are being prepared for IN-CORE integration.
Current projects
Seaside, Oregon (EQ + tsunami): end-to-end pipeline from hazard → damage → outcome prediction → bi-objective optimization for recovery/retrofit priorities; sensitivity to budgets, hazard scenarios, and model uncertainty.
Equity-aware planning: per-building fairness (normalizing by group size) to avoid “help where there are more assets” bias; stress-tests under budget cuts and shifting hazard intensity.
Critical network support: linking facility recovery to service areas (power/roads/shelter) so actions reflect system effects, not only individual buildings.
Data-lite risk scoring: tract-level ranking when only limited variables are available; separates exposure from residual harm to help target outreach and grants.
How we work
Methods: MILP with ε-constraint (and Benders when useful), Monte Carlo propagation from damage to outcomes, grouped out-of-sample validation, ablations (hazard-only vs damage-aware vs full graph).
Metrics: expected loss, downtime restored by time-T, number of impacted people, Theil index (inequity), calibration error, and frontier width between extreme policies.
Collaborations and students
We collaborate with civil engineers, data scientists, and local partners. Students in the lab learn problem formulation, Python/Gurobi, careful validation, and clear writing. If you are interested in hazard modeling, optimization, or decision support for real communities, email me with a short note on your interests and any relevant coursework or code samples.
OU Community Resilience Research Lab, School of Industrial and Systems Engineering, University of Oklahoma (Norman, Oklahoma, USA)
January 2021 – July 2025
Graduate Research Assistant
Co-advisors: Dr. Andres Gonzalez, and Dr. Charles Nicholson
The OU Community Resilience Research Laboratory, a part of NIST Center for Risk-Based Community Resilience Planning, is a research facility based at the University of Oklahoma that focuses on studying and enhancing community resilience in the face of natural and man-made disasters. Its interdisciplinary team of researchers work on developing and testing innovative strategies for disaster preparedness, response, and recovery.
Research Area:
Developing an optimization model for minimizing the loss of any community after any natural disaster. Analyzing the data of any community to find out the most important and relatable data for testing the optimization model.
Formulating efficient algorithms for solving a bi-level optimization problem to address the decisions of multiple decision-makers in the community during the impact of natural hazards.
As a citizen of Bangladesh, I have witnessed firsthand the devastating impacts of natural hazards, which are a regular occurrence in my country. From annual monsoon floods to cyclones that sweep across coastal areas, natural disasters affect millions of people, leaving behind destruction and hardship. One of the most vivid memories I have is the aftermath of Cyclone Sidr in 2007, which caused widespread damage, displacing thousands of families and severely impacting the nation’s infrastructure and economy. Events like this shaped my understanding of the urgent need for proactive solutions to manage and mitigate the risks associated with these disasters.
Bangladesh’s low-lying geography and dense population make it one of the most climate-vulnerable countries in the world. In response to these challenges, I developed a passion for finding ways to improve the resilience of communities, not just in Bangladesh but globally. My work focuses on leveraging optimization models and systems engineering to help communities better prepare for and recover from natural disasters. For example, I have worked on developing strategies to optimize the allocation of limited resources, such as flood barriers or emergency shelters, to minimize damage and reduce the number of displaced people.
One key aspect of my research is addressing the economic and social dimensions of disaster resilience. For instance, in many rural areas of Bangladesh, poor communities are disproportionately affected by floods, losing their homes, livelihoods, and access to essential services. By developing bi-level optimization models, I aim to help policymakers design mitigation plans that not only protect critical infrastructure but also prioritize the needs of the most vulnerable groups. These models integrate both community-level and individual-level decisions, ensuring that disaster response strategies are inclusive and equitable.
Ultimately, my passion for working on natural hazards is driven by a deep commitment to making a tangible difference. Through research that bridges engineering, data science, and social policy, I hope to contribute to the development of resilient systems that save lives, protect livelihoods, and foster long-term sustainability in the face of increasing environmental challenges. My journey from a country at the frontlines of climate change to working in this field globally has only strengthened my resolve to continue pursuing innovative solutions to one of the most pressing issues of our time.
The integration of optimization is essential in natural hazards research because it allows for efficient and informed decision-making, particularly when resources are limited, risks are high, and outcomes are uncertain. In disaster scenarios, there are often constraints on resources such as emergency shelters, medical supplies, and financial budgets. Optimization helps allocate these resources in a way that minimizes damage while maximizing the protection of affected communities. For instance, it can determine the best locations for flood barriers or emergency shelters, ensuring the greatest benefit with the least cost.
Another critical role of optimization is balancing competing objectives, such as minimizing economic losses, protecting human lives, and ensuring environmental sustainability. By using optimization techniques, decision-makers can explore trade-offs between these objectives and select solutions that are the most balanced and efficient. Optimization also helps address the uncertainty inherent in natural hazards by incorporating various disaster scenarios into models. This allows for the creation of robust strategies that can adapt to different possible outcomes, ensuring resilience even under changing conditions.
Moreover, optimization plays a crucial role in ensuring that disaster mitigation strategies are inclusive and equitable. Vulnerable populations, such as low-income households or remote communities, are often disproportionately affected by natural hazards. Through optimization, researchers can prioritize these communities in disaster response and mitigation efforts, ensuring that resources are distributed fairly. Lastly, optimization is vital for designing long-term resilience strategies that not only address immediate risks but also contribute to sustainable development. By integrating economic, social, and environmental factors, optimization helps create solutions that reduce risks and promote growth in vulnerable regions.
In summary, the integration of optimization in natural hazards research enables more effective, equitable, and sustainable decision-making, ultimately enhancing the resilience of communities in the face of disasters.
The future of research in natural hazards and optimization is evolving quickly, driven by the growing complexity of disasters and the increasing availability of data and advanced technology. As we look ahead, one major area of focus will be the integration of machine learning (ML) with optimization models. By combining the power of ML to process huge amounts of data—such as real-time weather information, satellite images, and even social media reports—researchers can develop more accurate predictions and smarter, faster responses to disasters. For example, optimization techniques can then take these insights and help create better evacuation plans or allocate emergency resources more effectively as conditions change on the ground.
Another key area of future research will be the shift toward decentralized decision-making. In many large-scale disasters, local authorities often need to act quickly and independently, especially when central coordination may be delayed. In response, future optimization models will likely focus on giving communities the tools to make smart, real-time decisions while still aligning with broader regional strategies. This allows for faster responses that are better tailored to the unique needs of different areas.
Handling uncertainty will remain a big challenge, and researchers will continue working on more robust models that can account for the unpredictable nature of disasters. The goal is to ensure that communities are not only prepared for known risks but also equipped to handle unexpected events. This kind of "robust optimization" will be crucial in making sure that strategies hold up even in the most unforeseen scenarios. Sustainability will also play a more prominent role in future research. With climate change increasing the frequency and severity of natural hazards, optimization models will need to find ways to balance immediate disaster response with long-term resilience and environmental sustainability. For instance, future solutions might focus on designing infrastructure that not only withstands disasters but also contributes to greener, more sustainable communities.
Lastly, we’ll likely see continued advancements in decision support systems. These systems will provide policymakers with real-time data and optimization tools to help them make smarter choices, both before a disaster strikes and in the critical recovery period afterward. The combination of machine learning, real-time data analysis, and multi-objective optimization will enable better, more adaptive decision-making that can save lives and resources.
In short, the future of research in this field will focus on using advanced technologies to create more adaptive, sustainable, and locally responsive solutions. By improving how we predict, prepare for, and respond to disasters, we can help communities become more resilient in the face of growing natural hazards.