Luiz F. G. dos Santos, Ph.D.

Machine Learning Engineer & AI Research Lead
LLM Systems · Distributed Training · Applied AI · Scalable ML Infrastructure

This repository contains the source code for my personal portfolio and academic research website at https://luizsantos.dev/.

I am an AI Research Lead and Machine Learning Engineer with a Ph.D. in Physics from The Catholic University of America (conducted with NASA Goddard Space Flight Center) and a John C. Mather Nobel Scholar. My work bridges theoretical physics, computational mathematics, and production deep learning systems.

• Delphos Labs Inc. — AI Researcher Lead directing core ML training infrastructure, distributed pretraining/fine-tuning (Composer, FlashAttention 3), custom tokenizers, and low-latency production inference (TensorRT-LLM, Triton, Kubernetes).
• Shell — Principal Data Scientist leading enterprise GenAI initiatives, RAG systems with contextual grounding, autonomous LLM agents, and deep learning on unstructured data.
• NASA GSFC & Frontier Development Lab — Space sciences machine learning research, including CNN-based autocalibration for the Solar Dynamics Observatory (published in Astronomy & Astrophysics) and deep neural networks for flux rope detection (Solar Physics).

• Hero & Identity — Overview of credentials, focus areas, and quick navigation.
• About Me — Research background, physics foundation, and technical philosophy.
• Experience Timeline — Detailed career history spanning Delphos Labs, Shell, CIRES/CU Boulder, FDL/SETI, and NASA Goddard, complete with accomplishments and technology stacks.
• Research & Publications — Peer-reviewed papers in Astronomy & Astrophysics (A&A), Solar Physics, and The Astrophysical Journal (ApJ), along with awards and honors.
• Education — Ph.D. in Physics (distinction), M.S. in Physics, and B.S. in Physics.
• Technical Expertise — Categorized competency grids covering LLM systems, GPU training & acceleration, inference serving, and data engineering.
• Featured Projects — Open-source starter kits, performance builds, and scientific templates.
• Contact — Direct email, LinkedIn, and research profile links.

• Frontend: React, React Router
• Styling: Sass (SCSS), Bootstrap, Custom Dark/Light Executive Theme
• Icons & Fonts: Font Awesome, Google Fonts (Inter)
• Deployment: GitHub Pages via gh-pages branch

• Node.js (v18+ recommended)
• npm

# Clone the repository
git clone https://github.com/luiz0992/luiz0992.github.io.git
cd luiz0992.github.io

# Install dependencies
npm install

# Start development server
npm start

Runs the application locally at http://localhost:3000.

# Build the optimized production bundle locally
npm run build

Deployment to GitHub Pages is fully automated via GitHub Actions (.github/workflows/deploy.yml). Any push to main automatically builds and deploys directly to https://luiz0992.github.io/.

This project is licensed under the MIT License.