📍 Chantilly, France | 🤖 SAP Business AI & BTP Solutions Architect @ VINCI Construction | 🚀 MSc in Artificial Intelligence @ Epitech

Ever since I got a machine with a Snapdragon X processor, my goal has been to run frontier LLM and VLM models natively using the NPU. This led me to Qualcomm's GenieX (formerly Nexa AI).

Because I love Rust, I decided to contribute by building the complete Rust bindings for their C API. Following discussions with the upstream maintainers, I extracted this work from my initial fork into a dedicated, standalone repository. The project supports automated Windows ARM64 dynamic linking out of the box and is published on crates.io.

• 📦 Repository: github.com/Maki-Grz/geniex-rs
• 🦀 Crates.io: crates.io/crates/geniex-rs
• 🔥 Official Website: aihub.qualcomm.com/geniex

Bringing high-performance developer tooling to SAP development. This extension provides first-class support for SAP Cloud Application Programming Model (CAP) CDS files directly within the Zed editor.

It leverages tree-sitter-cds for fast and accurate syntax highlighting alongside official @sap/cds-lsp integration to deliver diagnostics, completions, and Language Server capabilities natively inside Zed.

• 📦 Repository: github.com/Maki-Grz/sap-cds-zed
• ⚡ Zed Extension: zed.dev/extensions

Using Cline daily in my workflow, especially with SAP pushing for its adoption, I quickly ran into limitations: the existing SAP AI Core integration was incomplete, particularly around orchestration workflows and model allowlist handling.

To make it truly production-ready for daily use, I created a dedicated fork to fix these issues and enhance the provider. This branch resolves critical bugs, fully supports SAP's orchestration layer, and properly handles model allowlisting within the VS Code extension.

• 📦 My Fork: github.com/Maki-Grz/cline
• 🚀 Upstream Repository: github.com/cline/cline
• 💼 Official SAP AI Core: sap.com/products/artificial-intelligence

This solution automates the verification of physical deliveries against Purchase Orders. Users upload paper delivery receipts from SAP/Ariba purchase orders to the platform. Using Mistral OCR, the backend extracts the receipt data, compares it directly with the original SAP PO, generates a discrepancy score, and securely stores the results on SAP BTP / HANA for automated compliance tracking.

• 🖥️ Frontend: Svelte 5, Vite, SAP UI5 Web Components
• ⚙️ Backend: SAP CAP (Node.js), Mistral OCR Integration, SAP HANA

Patrimoine is an open-source, self-hostable, and sovereign personal wealth management platform. Built natively on the SAP Cloud Application Programming Model (CAP) and powered by Svelte 5 and SAP UI5 Web Components, it provides an elegant, security-first dashboard to map, automate, and simulate personal asset allocations.

Designed specifically for European citizens under French financial schemas, the application allows users to orchestrate incoming revenue splits into various bank accounts, savings books (Livret A, LDDS), company stock plans (PEG/PER), and investments (PEA, Assurance-Vie).

• 📦 Repository: github.com/Maki-Grz/patrimoine

• ⚡ sap-cds-zed (2026) - SAP CAP CDS language support for the Zed editor via tree-sitter-cds and @sap/cds-lsp.
• 🧩 extensions (2026) - Forked from zed-industries/extensions. Extensions ecosystem for the Zed editor.
• 🏦 patrimoine (2026) - Sovereign, SAP CAP-native personal financial wealth management application.
• 📖 sap-artificial-intelligence (2026) - Forked from SAP-docs/sap-artificial-intelligence. Markdown source for SAP AI documentation.

• 🦀 geniex-rs (2026) - Rust bindings for Qualcomm GenieX SDK (Snapdragon X / NPU, GPU, CPU).
• 🤖 Cline (2026) - Forked from cline/cline. Autonomous coding agent with SAP AI Core integration.
• 🦾 GenieX (2026) - Forked from qualcomm/GenieX. Run frontier LLMs and VLMs locally on Qualcomm devices.
• 🔗 rust-client (2026) - Forked from qdrant/rust-client. Official Rust client for Qdrant vector search engine.
• 🗂️ lumen-rag (2025) - Modular, database-agnostic RAG framework for Rust supporting MongoDB and Qdrant.

• ⚙️ gitbook2text (2025) - CLI tool to download GitBook pages and convert them to markdown and text.

• 🤖 smartmate (2026) - AI Procurement Copilot for S/4HANA invoice and delivery matching with Mistral OCR (Frontend / Backend).
• 🕸️ hydramesh (2026) - Next-generation, high-performance distributed VPN designed for enterprise-grade scalability.
• 📁 portfolio-v3 (2026) - Official portfolio (maxgrz.fr) built with Astro.
• 🪐 netanime (2026) - Anime streaming platform rewritten with microservices in TypeScript & Rust.
• 🍣 projet-sushi (2026) - Interactive Valentine's Day gift web application built with Svelte, Lenis, and GSAP.

• 🦀 aura-sdk (2026) - Rust bindings and CLI runner for Qualcomm Genie/QNN on Snapdragon X (superseded by geniex-rs).
• 🔬 ai-research-lab / ai-research-lab-rag (2025) - End-to-end AI research platform with high-performance Rust RAG engine (Groq, Qdrant).
• 💬 opencapai-chat-studio (2025) - Multi-model AI Chat Studio built with SAP CAP and SAPUI5 on SAP BTP.
• 👁️ RustSight (2026) - Face detection tool using SCRFD, ONNX Runtime, and OpenCV for Windows 11 ARM64.
• 🧠 mlp-mini-burn (2025) - Multi-Layer Perceptron (MLP) implementation using the Burn deep learning framework in Rust.
• 🕹️ treasurehunt-plugin (2025) - Custom Minecraft plugin for version 1.21.6 built for Floria server.
• 🗺️ dungeon-portfolio / dungeon-portfolio-api (2024-2025) - Minecraft Dungeons-inspired gamified portfolio (Vue/Nuxt + Actix-web/MongoDB).
• 📁 portfolio-v2 (2025) - Legacy portfolio built with Astro and Svelte.
• ⚙️ portfolio-azure-function (2024) - Serverless backend logic for portfolio management via Azure Functions.
• 🎨 star-wars-pixel (2023) - Interactive Star Wars-themed pixel art project using Canvas.
• 🧪 paladium-test (2023) - Design and integration tests for Paladium platform elements using SCSS.
• 🤖 bda-bot (2022) - Discord bot for Bataillon De l'Animation using YouTube & Genshin APIs.
• 🗺️ Znetwork (2021) - Educational project connecting channels across different Discord servers.
• 📉 prediction-tensorFlow (2021) - ML predictions directly in the browser using TensorFlow.js.
• 📷 webcam-tensorFlow (2021) - Real-time image recognition with TensorFlow.js using live webcam feed.
• 🐍 detect-face (2021) - Python project for face detection leveraging OpenCV.