ML4MS: Machine Learning for Materials Science
The Engineering Layer for Artificial Intelligence
🔭 The Mission
The gap between Academic AI (finding new crystals) and Industrial Engineering (Asset Integrity) is too wide.
ML4MS is a knowledge initiative dedicated to scouting emerging technologies and critiquing their viability for safety-critical operations. We move beyond “Chatbots” to build Engineering Agents that respect physics, adhere to global standards (NACE, API, ASTM), and operate with zero-trust architecture.
📝 Latest Research & Insights
From Guessing to Knowing: The Story of Machine Learning in Materials Science
Jan 04, 2026 | Topic: History, Fundamentals, Strategy
How We Went from Guessing to Knowing: The Story of Machine Learning in Materials Science
Beyond FEA: Solving the ‘Small Data’ Problem with Physics-Informed AI
Dec 29, 2025 | Topic: Simulation, Digital Twins, PINNs, Asset Integrity
Beyond FEA: Solving the “Small Data” Problem with Physics-Informed AI
800 Years in Weeks: Decoding Google DeepMind’s GNoME
Dec 25, 2025 | Topic: Discovery, DeepMind, Energy Transition, GNoME
800 Years in Weeks: Decoding Google DeepMind’s GNoME
From Atoms to Algorithms: How AI ‘Sees’ Materials
Dec 16, 2025 | Topic: Fundamentals, Graph Neural Networks, CGCNN
From Atoms to Algorithms: How AI ‘Sees’ Materials
The Death of the Lookup Table: Understanding MACE & M3GNet
Dec 06, 2025 | Topic: Simulation, Physics-AI, Asset Integrity
The Death of the Lookup Table: Understanding MACE, M3GNet, and the New Age of Alloy Simulation
Can We Trust AI with Sour Service Design?
Dec 05, 2025 | Topic: Asset Integrity, AI, Sour Service
The End of “Black Box” Engineering: A Blueprint for Safe AI in Asset Integrity A technical review of the ‘Crystalyse’ framework and a practical implementation of Provenance-Enforced AI for NACE MR0175 Compliance.
🧭 The Research Radar
Active investigations at the intersection of AI and Heavy Industry.
1. Accelerated Discovery & Generative Design
Investigating how Generative Adversarial Networks (GANs) and Diffusion Models are creating new alloy compositions.
2. Physics-Informed Machine Learning (PINNs)
Moving beyond traditional Finite Element Analysis (FEA). Exploring how neural networks can solve differential equations for corrosion rates 100x faster.
3. Computer Vision in Characterization
Automating the analysis of SEM (Scanning Electron Microscopy) and metallography images.
📡 Automated Intelligence Pipeline
Behind the scenes of ML4MS
To keep this initiative current, I maintain an automated n8n data pipeline that scans pre-print servers (arXiv) and major journals daily. It uses Natural Language Processing to filter noise and surface high-impact research.
👤 About the Editor
Ahmed Aburakhia Senior Materials & Project Engineer
I am an engineer first, and a data scientist second. My goal with ML4MS is to strip away the hype and find the tools that actually solve problems in the physical world.