Arash Behboodi
I am a research scientist and Director of Engineering at Qualcomm AI Research, working on the theoretical foundations of information and computational systems.
My interests include information theory, learning theory, machine learning for inverse problems, compressed sensing, geometric deep learning, differentiable simulation, neural reasoning and agentic AI.
Recent News and Highlights
Recent papers and research updates.
New paper: Explorer-Zero: Improving Self-Play for Long-Horizon Tasks with Decomposable Verifiable Rewards , NeurIPS Workshop CL4FMAgents, 2026
-
New paper: Enabling Dynamic Computation in Looped LMs
-
NeurIPS 2026: Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck
-
New paper: When Cloud Agents Meet Device Agents: Lessons from Hybrid Multi-Agent Systems
-
New paper: Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models and Hugging Face page
-
New paper on our NeurIPS 2025 demo: Efficient Reasoning on the Edge and Hugging Face page
-
New paper: LaneRoPE: Positional Encoding for Collaborative Parallel Reasoning and Generation
-
ACL Findings 2026: LUMINA: Long-horizon Understanding for Multi-turn Interactive Agents
-
New paper: Fundamental bounds on efficiency-confidence trade-off for transductive conformal prediction
Talks and Writing
Recent talks, blog posts, and activities.
-
Latest blog post: Bottleneck Theory of Post-training.
-
Keynote: Information Theoretic Perspective on Conformal Prediction at COPA 2025.