Hey, I'm Dan (people also sometimes call me Danny or Daniel). I’m a researcher working on agentic, multimodal, and efficient AI systems, among other things.

My work in AI began at MIT, leading some of the first efforts to study and design more energy-efficient AI systems (“Green AI”) on the algorithmic, hardware, and system levels. I previously worked at Microsoft as a senior research scientist on AI agents (e.g., Copilot), agentic benchmarks, and multimodal systems among other places like SandboxAQ (formerly GoogleX), Twitter, etc. In a past life, I've worked in government, trade, finance, drug discovery, gaming, etc.

Some areas my research focuses on include (but not necessarily limited to):

• Agentic AI: building, training, and evaluating agents that reason, plan, and act in realistic environments for general mulitmodal computer use [1][2][3], social simulation and social media platforms [4][5], agentic and AI safety + security [6], cybersecurity [7], and more.
• Efficient AI: develop new ways to holistically improve the efficiency of large-scale AI models and systems across hardware, systems, algorithms, training, and inference, spanning both theory and application.
  • Hardware, Systems, & Energy Efficiency: measuring and reducing compute/energy costs or large language model inference at scale [8], studying the effects of GPU power-capping system-wide, [9], benchmarking distributed training efficiency and resource utilization of various large-scale models, [10], and analyzing efficient AI infrastructure and systems [11].
  • Algorithmic & Model Efficiency: improving training and inference efficiency with new techniques for architecture searches [12], fast architecture ranking [13], efficient representation learning [14][15], model compression/sparsification, [16], quantization-aware training [17], etc.
• AI for Science: developing and applying AI to scientific workflows, spanning areas like quantum chemistry [18][19], quantum machine learning and computing [20][21], battery/materials research [22][23], etc.