AI Developer
Logibec, October 2017 – October 2022
Development of proof of concepts of machine learning and optimizations models in collaboration with product teams in order to solve business needs.
-
Developed a machine learning model able to predict the probability of stock out of a submitted order from an healthcare organization in order to optimize orders and limit late deliveries. Used Python, Databricks on Azure, Scikit-learn and Kedro.
-
Developed shared tools used by Logibec's products enabling features like single sign on, multi tenant, and SaaS on a cloud environment. Developed in Java with Spring Boot on Microsoft Azure.
-
Developed a machine learning pipeline supporting a new application predicting the risk of patients’ readmission before being discharged from the hospital. The application technology stack involved Apache Spark with Scala, Postgresql, Python, Flask, and Scikit-learn.
-
Involved in grant applications and coordination between Logibec, clients and academic researchers in order to get federal and provincial public funding for new machine learning projects relevant for healthcare organizations.
-
Managed machine learning interns from Mila who were responsible for analyzing data and developing new models relevant for the clients.
Machine learning specialist
Datacratic/iPerceptions, June 2015 – September 2017
Responsible for online real-time bidding advertising campaign performance made through Datacratic RTB-Opt product. Making sure campaign managers get the best ROI possible by monitoring machine learning models performance on key metrics (Click-Through Rate, Cost Per Click, Cost Per Acquisition).
-
Ensured continuous improvement of the campaign optimization system in order to fit campaign managers needs: enable optimization on small campaigns.
-
Improved the system C++ and Python technology stack in order to support new use cases and improve every day operations.
-
Developed a machine learning pipeline in Golang using Amazon AWS machine learning and Kubernetes to drive the new audience recognition technology.
Machine learning intern
Datacratic, September 2014 – January 2015
-
Research and development on application of neural networks for online real-time bidding advertising problems. Namely, predict accurately the probability of click by the user given an ad impression.
Research assistant
MILA, University of Montreal, September 2013 – September 2014
-
Help develop a Python tool to ease comparisons between machine learning models and enable research for new models for the facial landmarks recognition task.
-
Contributions to Pylearn2, a MILA library to ease machine learning model training workflow, in the form of refactoring the codebase and improve logging.