Antoine Mellerio

Development of a photo editing tools recommendation system for application users. Training and deploying a Machine Learning model that uses the user's past behaviors to predict the best tools for them. - Analysis of user view/click behavior - Development of a machine learning recommendation model trained with Metarank on view and click history - Implementation of recommendation API endpoints - A/B testing using Amplitude Stack : Python, Docker, Machine Learning, Metarank, Amplitude, MLflow

Development, optimisation and production deployment of GenAI apps for the internal use of the company (100 000 employees). • chatbot agent with multiple tools • advanced RAG • custom image generation pipelines Stack : python, FastAPI, Langchain, LangGraph, Docker, Azure, Bicep (Infra as a service), advanced CICD

Creation of an end-to-end pipeline including data collection using external APIs, data processing and dashboarding. https://www.malt.fr/profile/antoinemellerio

Accenta optimizes the heating systems of buildings, and pilot the heat distribution. Multiple AI algorithms have been developed based - among others - on energy consumption predictions. The objective of the internship is to conduct a survey on the latest forecasting technics and to test the ones that match the most to our context. We gave a specific focus on statistical models and on recurrent neural networks.

Forecasting project for a large retailer. More precisely, development of demand reconstruction models on historical data when out-of-stocks occured. This reconstruction yields to better predictions on the futur sales. The codes are written in python on AWS. Development and packaging of a recommendation system for consumer intelligence. The objective is to recommend products for each client of a large retailer.

Development of an automatic resume analysis module for the company's Machine Leaning Text Analyser software (MALTA). Module developed with Javascript leveraging OCR (Optical Character Recognition) and Machine Learning models.