00 AI Engineering · Howest · Belgium
WARRE
SNAET
Third-year AI Engineering student at Howest, Belgium. I spend most of my time getting ML models to work on local edge devices. Training on phones, inference in browsers. Right now that means a lot of Rust and the Burn framework.
SELECTED WORK
Semi-Supervised Learning in Rust
Research project exploring pseudo-labeling on embedded edge devices for plant disease classification. Built entirely in Rust using the Burn framework — 0.39ms inference, 2579 FPS, runs on iPhone.
Financial AI Agent
Conversational portfolio manager — ask questions, get advice, execute trades through natural language. RAG-powered with real market data, streaming responses, and full session persistence.
Event Chatbot — XPO Group
Full-stack RAG solution for event Q&A. .NET embedding service backed by Azure OpenAI and CosmosDB, Next.js analytics dashboard, and a Scrapy pipeline for live event ingestion.
WaterAdventure
Platform connecting swim coaches and parents to help toddlers (2–6) build water confidence. Features activity planning, medal-based achievement tracking, real-time chat, and multilingual support.
Dataset Query Socket App
Client-server socket application for dataset analysis with user authentication and moderator controls. Query LAPD arrest data, view popularity stats, and broadcast messages — all through a Qt GUI.