I am a Ph.D. student at Texas A&M University advised by Prof. Zhiwen Fan, and a Senior Research Engineer at Google DeepMind working on Gemini post-training and safety alignment.
Previously, I received my M.Sc. from Carnegie Mellon University and my B.Sc. from the University of Edinburgh, where I was advised by Prof. Hakan Bilen.
CoRL 2026
MoRE removes unwanted behavior modes from behavior-cloned policies, such as handing over a knife blade-first, with a short uncloning step and zero inference-time overhead.
NeurIPS 2026
Couples multi-agent LLM self-play with Bayesian optimization over a game's rule space to balance a civilization-style game automatically.
COLM 2026
Audits hidden behavioral entanglement among black-box LLMs and reweights verifier ensembles that wrongly assume their judges fail independently.
IROS 2026
LeanGate scores each frame's geometric utility before heavy foundation-model decoding, skipping over 90% of redundant frames in monocular SLAM.
arXiv 20262603.13528
Dream2Fix synthesizes photorealistic failures from successful demonstrations with a generative world model, producing 120K+ paired samples for learning executable recovery.
Findings of ACL 2025
Fine-tunes LLMs to simulate responses from 7,000+ real APIs, giving tool-learning benchmarks an environment that is stable, scalable and realistic.
Findings of ACL 2024
A virtual API server with caching and API simulators, plus solvable pass and win rates, for stable large-scale evaluation of tool use.
arXiv 20232311.10751
Introduces Agentic Process Automation: an LLM agent that builds workflows from human instructions and coordinates specialized agents at run time.
