Research

Reinforcement Learning:
I build sample-efficient and robust RL agents across model-free and model-based settings, with a focus on world models, efficient representations, and online adaptation for long-horizon decision-making. I also study robust multimodal RL and how generative/foundation models can strengthen RL in real-world applications (e.g., anomaly detection, recommendation). 

Reinforcement Learning projects →

Generative AI:
My work centers on generative representation learning, especially VAEs and their modern variants (hierarchical VAEs, VQ-VAEs), to learn compact latent spaces for compression, robustness, world models, and anomaly detection. I also work on GAN-based robustness, diffusion models, and LLM-guided reward shaping in RL. 

Generative AI projects →

Applied Machine Learning:
I develop applied ML systems in geospatial, computer vision, and health domains, including GeoAI for drainage-structure detection from LiDAR DEMs and multispectral pedestrian detection for nighttime safety with robustness/bias analysis. I also use interpretable ML to study subgroup-specific drivers of quality of life in autistic adults in an NIH R01/CBPR collaboration. 

Applied Machine Learning projects →