Kocay Migo

Project: Implementation of a monitoring tool for machine learning models in production to track their behavior over time.

• Monitoring input data for data quality (Completeness, compliance...).

• Monitoring predictions (Target and explanatory variable drift, distribution, distance calculation...).

• Monitoring model performance.

• Implementation of a package that allows dynamic exploration of training data (IPwidget, plotly).

. BigQuery query for data retrieval, creation, and transformation

. Creation of Airflow DAGs.

Project: Solidify a cross-functional view of company data on a Cloud technology. This vision should align the company on indicators and definitions, and provide dashboards for targeted and strategic company management.

• Design a persona per department (product, marketing, sales, actuarial). For each department, identify the role, organization, decisions to be made, desired data transformations, analyses performed, and encountered challenges.

• Design of various activity monitoring metrics per department presented in Power BI.

• Creation of a schema that consolidates needs, personas, and data to better investigate data in source databases.

• Creation of a conceptual data model.

• Data Lineage of business data by identifying sources and data to retrieve, creating traceability of the data lifecycle from different databases, and determining data types (sensitive data, personnel data, security data...) to implement a data governance strategy.

• Implementation of a centralized functional and technical Cloud architecture on Azure Cloud.

• Creation of data pipelines using Azure Data Factory.

. Data transformation in Python.

Project: Development of a user interface for verbatim analysis. The application aims to help users analyze all types of text (verbatims, comments, reviews, tweets...). The user uploads a file containing verbatims, and the application provides several analysis modules (Word Cloud, Sentiment Analysis, Topic Modeling, Text Classification, Clustering, Knowledge Graph).

• Text processing and treatment

• Development of Word Clouds based on requested filters

• Development of a sentiment analysis model

• Development of a topic modeling model

• Clustering of negative verbatims

• Initial development of a knowledge graph model

• Creation of a PostgreSQL database and development of the application's backend using Flask with Python

• API security using OAuth2 protocol

• Monitoring of the CI/CD pipeline for production deployment on Gitlab

• Teamwork with DevOps and front-end developers.