Mohammad Askari

About

I am a Ph.D. candidate in Civil Engineering at UCLA, advised by Professor Ertugrul Taciroglu, and an M.S. candidate in Statistics. I build machine learning systems that estimate natural-hazard risk at regional scale, from raw data to the losses communities can expect.

My research projects include:

• CHEERSafe: I lead development of this parallelized Python library, which estimates hurricane wind and flood losses for 1.15 million single-family homes across North Carolina for the NSF CoPe-CHEER project.
• AI-driven building inventories: fusing FEMA, Microsoft, NSI, and parcel data with vision transformers and gradient-boosting imputation to describe 5.65 million buildings in North Carolina.
• Post-earthquake damage assessment: RAPID-A, an uncertainty-aware deep ensemble that maps building damage from satellite images of the 2023 Türkiye earthquakes.
• Insurance claims modeling: co-developed XGBoost models that predict which buildings file hurricane wind claims, built on 736,000 building–hurricane observations.
• Ground-motion selection: a GPU-resident JAX search that selects earthquake records more than 2,000 times faster than the previous OpenCL tool.

Outside my research, I build open-source tools and study large language models. I created Overleaf Web MCP, which lets AI assistants read, edit, and compile Overleaf projects, and I have studied how proper scoring rules reduce LLM overconfidence.

My goal is to use AI-powered techniques to address real-world problems such as natural disasters and the challenges they put on the community. I am increasingly enthusiastic about AI for Good and cannot wait to see how it will enhance our understanding of the Earth and the universe.

CV

Projects

Under Construction! :)

Blog

Contact

• Email
• LinkedIn
• GitHub

Elements

i = 0;

while (!deck.isInOrder()) {
    print 'Iteration ' + i;
    deck.shuffle();
    i++;
}

print 'It took ' + i + ' iterations to sort the deck.';