Diadie Sow

Senior Data Scientist / Machine Learning Engineer – DSCP Product

Designed and deployed automated demand forecasting models for multiple international clients. Built scalable batch inference pipelines orchestrated with AWS Step Functions with scheduled retraining strategies. Implemented data drift detection to ensure model performance over time. Delivered production‑grade ML solutions integrated into a multi‑tenant supply chain planning platform.

Data Scientist – Industrial AI Projects

Developed end‑to‑end machine learning systems from data ingestion to deployment. Delivered customer segmentation and predictive analytics solutions with direct business impact. Industrialized model training and inference workflows using containerized environments and cloud infrastructure.

Mentor data science student. This include explain them the differentcourses and mentor their different projects around Machine learning, deeplearning and their applications on:predictive modeling, customersegmentation, natural language processing, image processing. someexemple of projects:

. Using machine learning models for predicting consumption andemissions of buildings not intended for residential use in the City of Seattlein the USA

. Provide a Brazilian e-commerce site segmentation of its customersusable for their communication and marketing campaigns using Clustering

. Automatically classify consumer goods: Implementation of a productclassification engine through a text and/or image description. Use of NLPtype techniques to extract relevant information through a description: Bagof words, Tf-idf, Word2vect, FastText….use algorithms to extract featuresfrom images: CNN, transfer learning...

. Implementation of a “credit scoring” tool for a local bank to calculate theprobability that a customer repays their credit then classifies the requestas credit granted or refused. Use different classification models thencompare them, carefully manage the case of imbalance on the 2 classesin the training data, implementation of a cost function adapted to the creditscoring context. Set up an API and interactive dashboard in the cloud.

Technical stack: Jupyter notebook-Python, github, git, aws, Heroku

Subject: Robust optimization with recourse for short-term productionmanagement

In the short term, the problem of planning the production of each means(Unit-commitment problem) aims to calculate the production programwhich satisfies the equilibrium offer-demand at a minimum cost. Theseproblems are often considered in a deterministic framework, i.e. thedemand and availability of the means of production are seen with certainty.However, the uncertainty is indeed present, so we place ourselves withinthe framework of robust optimization to try to construct programs facingthe uncertainty, such as temperature, cloudiness, affecting the demand. Inthis project, I proposed a two-step m