ResearchI design memristive associative memory hardware for AI inference and reasoning, working across the device–circuit–architecture stack to make associative computation a first-class primitive in modern AI systems. My work spans three thrusts: Memristive SubstratesDevice-level cells, volumetric arrays, and characterization for in-memory computing. Projects: 3D Cellular Neural Networks for volumetric computing (with Chua, Williams, Xia); RRAM crossbar device characterization with HPE Labs. Associative Memory CircuitsCAM cells, arrays, and architectures for attention, retrieval, and similarity matching at hardware scale. Projects: CAMformer (associative memory for attention), MonoSparse-CAM (tree models in CAM), NP-CAM (scalable CAM architectures), and a comprehensive review of 37 CAM cell designs in IEEE TCAS-I. Hardware for Inference and ReasoningAlgorithm–hardware co-design for the workloads associative memory enables — attention, tree search, neuro-symbolic reasoning, and test-time compute. Projects: IMC-MCTS (in-memory tree search), Neuro-Symbolic Policy (rule lookup in CAM), Hamming Attention Distillation (binarized attention). |