Hi, I'm Noé Le Yhuelic
My expertise spans the full AI pipeline, from understanding how models are built to effectively using them to create innovative, real-world solutions. I am particularly interested in leveraging state-of-the-art models through agent orchestration, MCP servers, and system-level integrations. I enjoy focusing on how these models can be combined, deployed, and adapted to solve concrete problems and build scalable, impactful products.
My Skills
Featured Projects
Voice-Interactive Conversational AI Agent
An intelligent, multilingual voice assistant system designed for restaurant operations. Built with LangChain, it handles reservations, orders, and general inquiries through voice or phone calls using advanced AI agents, RAG-based knowledge retrieval, and speech processing.
Explainable AI Interface
A unified web application for multi-modal AI classification with explainable AI (XAI) techniques. This project integrates deepfake audio detection and lung cancer detection into a single platform with full support for LIME, SHAP, and Grad-CAM visualizations.
Hackathon ESILV x IBM
This project was developed for an IBM Hackathon focused on building a Business Intelligence pipeline with AI-powered question answering capabilities. The system combines a traditional help center database with semantic search using embeddings and OpenAI's GPT models to provide intelligent responses to user queries.
RAG for Job Offers
This project implements a Retrieval-Augmented Generation (RAG) workflow that retrieves, processes, and analyzes job offers from the Civiweb API. It leverages OpenAI’s API to enhance or summarize the collected data, enabling smarter insights or automated Q&A over job postings.
SatGen-CV
This is SatGen-CV, a Pix2Pix-based system for generating realistic satellite imagery from semantic segmentation masks that contain roads and buildings. Our approach addresses the practical challenge of generating synthetic largescale aerial imagery by combining a U-Net generator with PatchGAN discriminator and multi-layer perceptual loss.
