Building the next generation of generative AI that people can control, align, and personalize—from long-horizon video to interactive worlds. I lead GEMLAB at Virginia Tech.
Generative models can imagine almost anything: I make them controllable.
I am an Assistant Professor of CS at Virginia Tech, where I lead 💎 GEMLAB, the Generative Modeling Lab. My research develops generative AI that people can control, align with their goals, and personalize to their needs, with a particular focus on video generation and interactive world models. I investigate how model representations, memory, and feedback can make these systems more efficient, responsive, and capable of following human intent.
Since I joined Virginia Tech in Fall 2023, my lab has produced 25+ papers at top conferences including CVPR, NeurIPS, and ICCV/ECCV. My research is supported by an NSF CAREER award on generative cinematography and industry support from companies including Google Research, Adobe Research, Snapchat, and Qualcomm.
Prior to joining Virginia Tech, I received my Ph.D. in Computer Science from Purdue as a Fulbright Fellow and was a postdoctoral researcher at the MIT Media Lab. My long-term goal is to advance the foundations of generative AI, enabling models that remain coherent over long horizons, respond reliably to human guidance, and adapt to individual goals.
I study how generative models represent content, motion, and temporal context, and turn these insights into efficient methods for control. My work spans interpretable representations, training-free interventions, and redesigned memory mechanisms for controllable images, long-horizon video, and interactive worlds.
I develop reward- and preference-guided methods, including reinforcement learning, that steer generation toward desired outcomes such as diversity, physical plausibility, and visual quality. These methods guide fixed generators through their inputs or adapt the generative models themselves.
I personalize generative models to both what people want to depict and how they want it to look. My work spans concept customization and individual preference learning, enabling generation that reflects a user's subjects, visual tastes, and creative goals.
Find papers, code, datasets, demos, and more on our website.
A selection of recent peer-reviewed work.
Recent papers, awards, and recognition from the lab.
Spanning long-horizon video synthesis, physically aligned video generation, and controllable world models.
Honored to receive the NSF's CAREER award, supporting our research and education in controllable, accessible generative AI. Award details ↗
Recognized for outstanding service on the program committee for #CVPR2026.
The lab will present four papers at CVPR 2026, including ∞-RoPE for infinite, action-controllable video and DPP-GRPO for diverse video generation.
For concept-level interpretability in diffusion models, at the CVPR 2025 Visual Concepts Workshop.
Spanning training-free dynamic view synthesis to personalizing content, including a Spotlight selection.
Controllable and personalizable image editing and generation.
Including CLoRA, selected as a Highlight.
Recent keynotes and invited talks on generative AI.
Below are some of the courses I’ve recently taught.
Interested in collaboration, prospective graduate study, or industry partnership? I'd love to hear from you.