Hi, I’m Lumina 🖤

I build intelligent systems to explore the frontiers of machine reasoning. After a strategic pivot from a Biochemistry focus at UCLA, I now dedicate my time to fundamental AI research, systems architecture, and speculative fiction.

Featured Work

Featured Publication

Epsilon: A Transformer with Adaptive Computation and Quantized Attention

The quadratic complexity of the standard Transformer presents a significant bottleneck. To address this, we introduce EPSILON, a novel architecture designed for high efficiency, training stability, and interpretability. It synthesizes adaptive computation, efficient quantized attention (HQSA), and architectural safeguards to significantly outperform a comparable baseline in both accuracy and speed.

Read the Paper (PDF) View Code on GitHub

Featured System: Project Prometheus

Mission Success: A Novel Molecule Discovered

Project Prometheus, an autonomous AI scientist, successfully designed a novel molecule with a Composite Score of 9.661, significantly outperforming the baseline and V2.0 Legacy. The system replicates the full scientific method from research to whitepaper to accelerate autonomous discovery.

View on GitHub Read the AI's Report

← →

Featured Research & Engineering

• Epsilon: A novel Transformer architecture with adaptive computation and quantized attention, designed for high efficiency and training stability.
• Prometheus: An autonomous AI scientist that grounds LLM reasoning in physics-based simulation to design and validate novel molecules.
• CodeTag: A professional-grade toolkit for codebase analysis and LLM context packing, featuring a TUI, CLI, and integrated security scanners.

Reading List

• A Thousand Plateaus — Gilles Deleuze & Félix Guattari
• Gödel, Escher, Bach — Douglas Hofstadter
• Solaris — Stanisław Lem

My Curriculum Vitae

For a detailed, one-page summary of my research, projects, and professional experience, please download the latest version of my CV.

Download CV (PDF)