Moving to Matrices: Multiple Linear Regression in Vanilla Rust
Stepping up from simple lines to multi-dimensional planes. A deep dive into Multiple Linear Regression, Feature Scaling, and Matrix math in Rust.
AI Research Lab
End-to-End AI Research Journey
Models
Fields
Mechanics
The Objective
This journal documents my end-to-end learning journey across Artificial Intelligence. The goal is to build a profound understanding of neural network architectures, from fundamental Multi-Layer Perceptrons to state-of-the-art Large Language Models.
Machine learning is evolving from academic theory into the foundational infrastructure of software. This website serves as a public index of my research, mathematical derivations, and code implementations.
Rather than treating models as black boxes, my approach focuses on understanding the underlying mechanics. This includes studying the optimization functions in , exploring how can rival larger counterparts through data quality, and dissecting the statistics behind .
By implementing these systems, leveraging techniques like , and analyzing research papers, I am bridging the gap between theoretical concepts and practical deployments.
My latest findings, implementation notes, and insights are chronologically documented below:
Latest Publications
Concrete AI projects
delivered in production
"Integrated LLM capabilities directly into SAP CAP applications via SAP AI Core, enabling context-aware responses grounded in live business data without leaving the existing ERP workflow."
"Developed a Joule Skill for automated Purchase Order creation within the Vinci Construction environment, reducing manual entry time and surfacing AI-assisted suggestions at the point of procurement."
Built a Joule Skill for Work Breakdown Structure management inside Vinci Construction's SAP landscape. The skill lets project managers query, create and update WBS elements through natural language directly in the SAP Build Work Zone interface - no transaction codes, no training overhead. The entire flow, from intent recognition through SAP AI Core to the CAP service layer, was designed and deployed end-to-end. A concrete example of AI meeting enterprise-grade production constraints.
"Development of a machine learning model in linear regression on data from data.gouv.fr (fr-esr-insertion_professionnelle-master) for salary prediction."
"Rust-based learning project demonstrating a Multi-Layer Perceptron (MLP) implementation using the Burn deep learning framework. This repository is designed as a hands-on base for exploring neural networks, data preprocessing, training, and model evaluation"