Maxime Benoit

RapMinerz is the first data driven media about french Hip-Hop, with more than 50k followers online Our ambition : Use the latest technological innovations to offer a community of enthusiasts an in-depth and unprecedented vision of french speaking Rap. We have aggregated one of the largest French Rap databases and then developed our own analysis technology using the latest open-source algorithms. Data and statistics do not have a very good reputation with the general public. We want to go beyond the raw and cold use of data to deal with varied and entertaining subjects. We reflect on our formats so that they go beyond simple data visualization: interviews, freestyles, videos, web applications, etc. Follow us on > instagram @rapminerz > youtube @rapminerz > twitter @rapminerz > tiktok @rapminerz

Shift Technology works with the world's leading insurance carriers, delivering superior AI-native fraud detection and claim handling approaches with a SaaS model. As a member of the fraud data science team, and working alongside our technical experts, my role is key to improve and roll-out our solution, meaning: > Launching and administrating my client's SaaS production with constant improvements, directly dealing with clients needs every week > Presenting demos and training clients to use our tools > R&D for product development and innovation : currently topic manager of « ML Integration & Scoring » Our stack : Microsoft SQL - Elasticsearch - .NET Framework (C⛏ on JetBrains Rider - ML.NET - SharpLearning) - Python - JetBrains TeamCity - Octopus Deploy - React.js

Study of Big Data and AI applications to Streaming Delivery > Data Science Lab Update : Cluster configuration for Big Data : Hadoop/Spark and NoSQL Databases Nvidia Geforce GTX 1080Ti GPU configuration for Deep Learning applications > Content Delivery Network (CDN) Log Mining for KPIs ingestion : Study, extraction and storage of KPI time series using Scala and Spark Network dynamic visualisation with Mapbox (GeoJSON, HTML-CSS-JS) > Anomaly Detection over CDN KPIs at network scale Time Series analysis : correlation, seasonal-trend decomposition procedure based on Loess Anomaly Detection with density-based spatial clustering of applications with noise (DBSCAN) Pattern recognition using Deep Neural Networks on GPU Study of Hierarchical Temporal Memory (HTM)