INIT ARCHITECTURE:Arash Nikzad
"Synthesizing Biological Intelligence with Machine Architectures"
I specialize in the intersection of computational neuroscience and deep learning, building autonomous systems inspired by the Thousand Brains Theory to achieve robust, hierarchical world modeling.
About Me
Arash Nikzad
RESEARCH_IDENTIFIER: 8842-X
My research trajectory is centered on the formalization of neocortical principles into scalable machine learning frameworks. Deeply influenced by the Thousand Brains Theory of Intelligence, I develop systems that transition from static pattern recognition to dynamic, sensory-motor world modeling.
Currently, I am investigating how sparse distributed representations, grid-cell based navigation, and hierarchical predictive coding can be integrated into modern RL and generative agents. My goal is to build intelligence that doesn't just process data, but understands the underlying structural causal models of its environment.
Technical Skills
Theoretical AI/ML
Computational Neuroscience
Core Engineering
Research Frameworks
Mathematical Foundations
Deployment Systems
Featured Projects
Cortical-Net v1.0
A large-scale implementation of a thousand-brains inspired architecture, achieving superior generalization in 3D object recognition through sensory-motor integration.
VQ-VAE + HMM Sequence Engine
A hybrid probabilistic-generative model for unsupervised discovery of temporal hierarchies in high-dimensional video streams.
Active-Sensation RL Agent
Autonomous agent utilizing grid-cell representations for efficient exploration and mapping of non-Euclidean environments.
Distributed Neocortical Simulator
High-performance C++ simulator for studying the synchronization and voting dynamics between thousands of simulated cortical columns.
Research & Vision
Developing a unified framework for intelligence based on the distributed, voting-based architecture of the neocortex.
Cortical Micro-Architectures
Modeling individual cortical columns as independent, sensory-motor learning units capable of object-centric representation.
SDR & Memory Systems
Utilizing Sparse Distributed Representations (SDRs) to achieve high-capacity, robust, and biologically plausible memory.
Hierarchical World Models
Building multi-level predictive systems that capture environmental regularities through active inference and grid cell mapping.
Biomimetic RL Agents
Engineering agents that utilize active sensation and internal coordinate systems for complex navigation and manipulation.
Professional Journey
Lead Research Engineer
2023 - PRESENTFrontier Neuroscience Lab
Architecting large-scale cortical models and optimizing neural simulation kernels for distributed GPU clusters.
Graduate Researcher
2021 - 2023Advanced Intelligence Institute
Developed hierarchical representation learning algorithms and contributed to the peer-reviewed research on sensory-motor integration.
Get In Touch
Ready for Collaboration
I am actively seeking research collaborations, technical discussions, and opportunities at the frontier of AI and neuroscience.