Apr 2024 – Mar 2025 · 11 mos · Remote

• Leading our core ads modeling efforts from retrieval to serving. Working on the typical large scale setups with sophisticated sequential architectures and GPU accelerations.

Jun 2017 – Apr 2024 · 6 yrs 10 mos · New York City Metropolitan Area · Remote

• At Instagram my focus areas are recommendations, ranking, content understanding and spam detection. I wear multiple hats ranging from individual contributor to lead researcher. My ever growing team has driven a meaningful amount of revenue for the company over the years. I am directly responsible for setting directions for ML at Instagram and researching/implementing the next generation of recommendation systems.
• I have mentored many scientists and engineers and have been involved in quite a few public conferences and talks with the rest of the ML community.

May 2015 – Jun 2017 · 2 yrs 1 mo · New York City Metropolitan Area

• I was in the Ads & Commerce pillar mostly focussed on the DFP Platform (Google Ads).
• Amongst my areas of focus was the reporting pipe - a high scale data analytics system where I made various engineering upgrades and improvements from full stack engineering to statistical fundamentals.
• My other center of interest was insights and yield optimization where I got involved with advanced auction theory and built inventory management systems.
• I gave various talks during my tenure around Ads ecosystem, auction design, game theory and hypercube analytics.

Jan 2014 – May 2015 · 1 yr 4 mos · New York City Metropolitan Area

• I worked on a range of projects from ad campaign sentiment analysis using NLP to anomaly detection.
• I spent most of my time on campaign effectiveness and yield optimization. I've integrated with various ad serving systems and built real time statistical models for large data streams. I also wrote clustering methods to model attributions for ads ROI to major events.

Sep 2013 – Jan 2014 · 4 mos · Brussels

• Developed a NodeJS deploy application.
• Worked with packer to provision an EC2 Ami.
• Handled AWS CloudFormation API through a NodeJS aws-sdk.
• Enhanced deploy flow through AWS Route 53 api.
• Used Express Framework / Handlebars with NodeJS.
• Enhanced ansible playbook to build AWS EC2 Instances.
• Built server architecture using amongst other things AWS S3, Glacier, RDS and MongoDB

Jan 2013 – Sep 2013 · 8 mos · Brussels Be

• Developed a Ruby on Rails project using MVC architecture.
• Created real time reporting by enhancing and using the open source fnordmetric (https://github.com/paulasmuth/fnordmetric)
• Ruby plugin
• Worked under the Laravel PHP framework using artisan.
• Implemented real time analytic tools going through the EPO third party data.
• Configured web architecture using chef from Opscode.
• Used a lot of common gems such as devise, omniauth,simple form etc…

Jan 2010 – Jan 2013 · 3 yrs · New York

• Created an open source project ‘Alternator’ to mimic AWS DynamoDB locally (https://github.com/mboudreau/Alternator/)
• Developed a maven project using Java Spring framework with MVC architecture.
• Implemented automatic backup of our data using AWS S3, EC2 and Elastic Map Reduce/Hadoop.
• Invented a way to have geo localization in data using a combination of DynamoDB and Cloud Search.
• Created daily statistical reports of our data, that helped noticed trends within the use of our product.
• Participated for 3 months to the Syracuse StartFast accelerator program.
• Designed some front-end pages using twitter bootstrap, backbone.js and jQuery.
• Researched and used several different third party APIs like Google Places, Google Calendar, Facebook, Twitter, Foursquare and
• Meetup.

Sep 2009 – Jun 2010 · 9 mos · Greater Paris Metropolitan Region

• Proved the effect of estimated parameters in backtesting Value-at-Risk models
• Corrected estimation induced error in Value-at-Risk backtesting models by taking into account estimated parameters and modifying the
• backtesting procedure improved most classic backtesting procedures by 20-30%.