I am a professional mathematician and AI researcher holding a Ph.D. from the Department of Mathematics at Korea University. I characterize myself as an inventor who views artificial intelligence not merely as an engineering discipline, but as a profound geometric phenomenon. My research is driven by a core conviction: that the most challenging bottlenecks in deep learning are, at their root, unresolved problems in geometry and topology.

My work bridges classical algebraic geometry with state-of-the-art AI architectures, replacing fragile empirical heuristics with absolute mathematical guarantees. As a founding and senior member of the AIMLK lab, I have conceptualized and led diverse interdisciplinary projects across time-series signal processing (AI4Science), NLP, and computer vision. My foundational research focuses on creating principled reparameterization and extension paradigms, including:

• Catalyst & IPPRO: Lossless structured pruning and scale-invariant importance metrics leveraging algebraic varieties and projective geometry ($\mathbb{RP}^N$).
• One-Point Contraction (OPC) & FM-recovery: Irreversible machine unlearning and diagnostic attacks via the intrinsic geometry of softmax preimages.
• Autokinematics: Bypassing topological barriers in mechanical linkage design using algebraic topology and Mixture-of-Experts (MoE) architectures. -Bypassing: Resolving stationary point problem in neural network training, by characterizing given landscape as algebraic variety in perspective of high dimensional space

Moving forward, my mission is to establish algebraic geometry as a foundational language for modern AI—turning empirical observations into universal geometric truths.