Romain Mondon-Cancel

Software Engineer, Data Science

Datadog – Paris, France

• Working in the Watchdog Metrics team, responsible for automatically detecting anomalies in customers' infrastructure and integration metrics, distributed over thousands of Python-based workers deployed on multi-cloud (AWS, GCP, and Azure). The service mesh relies on Protobuf + gRPC.
• Migrated to a release train deployment tool across multiple teams (totaling ~25 persons), including mentoring, technical documentation, and development workflow transition.
• Developed a MLOps monitoring proof of concept to track the performance of a model from training to production.
• Developed and evangelized good development practices with extending support for type hints, developing unit testing frameworks, and implementing productivity tools like code coverage difference.
• Presented papers on weekly paper reading groups to share state of the art and interesting breakthroughs in the Data Science / Machine Learning ecosystem.

Senior Machine Learning Engineer

Ubisoft – San Francisco

• Developed a TensorFlow and Keras-based deep recommending model for Rocksmith+ to personalize in-game content based on user behavior (millions of users, thousands of contents).
• Designed, built, deployed, and monitored a back-end Kubernetes application on GCP to expose a REST FastAPI-based service able to train our models and return predictions.
• Set up the DevOps for this application, enforcing a semantic versioning-based lifecycle and developing CI-CD operations to automate testing, documentation and releases using GitLab.
• Designed, trained, improved, deployed and tested a logistic regression model based on player gaming data to improve CRM targeting by up to 5%.

CMK Data Scientist

Ubisoft – Paris area

• Data processing and analysis of the behavior of players on multiple AAA and mobile games (+10 millions users): created user segmentations for customer understanding and CRM targeting.
• Animated formations on Data Analysis, Statistics, Machine Learning and Python Programming.

Junior Data Scientist

Ubisoft – San Francisco

• Data processing and analysis of the behavior of players on multiple AAA and mobile games (millions of users): created user segmentations for customer understanding and CRM targeting.
• Developed a front-end framework based on R-Shiny to quickly deploy web applications combining typical data and statistical visualization and analyses.
• Developed a NLP model based on TF-IDF to cluster documents around similar topics to partially automate survey data analysis. Deployed behind a job queue to process the payloads asynchronously.

Research Engineer Intern

Argonne National Laboratory – Chicago area

• Developed an innovative Markov Chains-based algorithm based on that profile to generate realistic speed profiles on any itinerary, significantly improving the quality of carbon dioxyde emissions simulations and engine performance.

MITx

Graduate-level courses – EdX

• Fundamentals of Statistics: passed with 100%.
• Machine Learning with Python – From Linear Models to Deep Learning: passed with 100%.

Mines ParisTech

M.S. Mechanical and Energy Engineering – Paris area

Major in Energy Physics.

École polytechnique

M.S. Engineering and Business Administration – Paris area

Minors in Mathematics, Physics, and Computer Science.

Resume website

https://github.com/skasch/skaschcv

Static website written in Flask and Python, with YAML-based templating for easily editable content and French-English localization. The resume is containerized and deployed on a Docker container.

HeroRotation Generator

https://github.com/herotc/hero-rotation-generator

Wrote a SimulationCraft to Lua transpiler in Python. SimulationCraft is a C++ simulator for World of Warcraft with its own scripting language to define character strategies. My project transpiles a profile written in this scripting language into Lua code, to be able to run the logic inside a World of Warcraft add-on.