Smbat Shakhpazyan

I am a Machine Learning and Deep Learning researcher with over 3 years of experience building and optimizing high-performance models. I specialize in moving beyond "black-box" implementations by applying rigorous mathematical intuition to solve complex tasks. My work bridges the gap between high-level academic research and production-ready applications.

Core Expertise
Deep Learning Research: Advanced implementation of CNNs, Transformers, and Graph Neural Networks (GNNs).

Mathematical Modeling: Expert proficiency in Linear Algebra, Calculus, and Probability used to optimize model convergence (e.g., EM Algorithm, Backpropagation).

Supervised & Unsupervised Learning: Custom development of SVMs, Ensemble methods, Softmax Regression, and Clustering models.

Full-Stack AI Deployment: Developing end-to-end systems with Django and Next.js to bring AI models to life.

Academic Training
My technical foundation is built on the rigorous curriculum of Stanford University and my current degree program:

Stanford CS229 (Machine Learning): Comprehensive mastery of statistical pattern recognition, generative learning, and reinforcement learning.

Stanford CS109 (Probability for Computer Scientists): Deep expertise in probabilistic models, parameter estimation, and uncertainty theory.

Yerevan State University: Currently pursuing a degree in the Faculty of Informatics and Applied Mathematics, focusing on Graph Theory and Discrete Mathematics.

Featured Projects
Next Academy: Designing and building a hybrid Online Judge and Learning Management System using Next.js and Django.

Spatiotemporal Analysis: Researching sentiment propagation on global supply chain graphs using GCN layers and adjacency matrix logic.

Optimization Systems: Building and refining custom AI-powered tools, including handwriting recognition and video remix applications.