Phil Gaudreau Email & Phone Number

San Francisco, California, US

Sunnyvale, CA, US

• Taking on the role of Tech Lead for scaling Project Y, a cutting-edge initiative focused on making Double Machine Learning (DML) causal inference analyses more accessible, particularly in the context of conversion.
• I lead a cross-functional team overseeing the training of numerous neural network forecasting models in PyTorch for LinkedIn’s capacity planning, surpassing baseline models by over 15% in accuracy. Pioneered novel.
• My latest model used in production resulted in $13.1 million saved compared to the previous approaches.
• Led scaling of LinkedIn’s forecasting and anomaly detection hyper-parameter tuning framework: Greykite-on-Spark, boosting speeds by 3811%, saving an estimated 360 engineering hours. Enabled versatile forecasting.
• Designed Kafka cluster and topic growth model pipeline, predicting Kafka metrics 18 months ahead for capacity planning. Operationalized Kafka anomaly detection pipeline (90% precision), monitoring 50K topics daily.
• I lead a cross-functional team overseeing the design of a standardized Media Mix Modeling framework at LinkedIn, which allows us to measure the true impact of all our marketing channels (not just online channels) and.

Renningen, Baden-Württemberg, DE

• Led the development of a Bosch-wide forecasting tool that delivers scalable and automated forecasts for various time series. The tool’s goal was to enhance forecasting accuracy across the organization. The solution was.
• Responsible for technical development, project planning, and stakeholder management.
• Delivered tested and reusable code, contributed to code reviews, and designed a continuous integration framework for our pipeline.
• Coordinated the migration of the forecasting pipeline to Microsoft Azure cloud services, working closely with the Azure team to ensure a smooth transition.
• Developed a meta-learning Random Forest model to reduce computational costs by 5x and improve overall accuracy over the existing method. Took the initiative to research and implement this model, resulting in.
• Effectively communicated results and ideas to stakeholders, including customers, managers, and key decision-makers, to ensure a clear understanding of our work and its impact.

Renningen, Baden-Württemberg, DE

· Execute complex data science / data engineering projects that have a significant impact on Bosch global business.· Share knowledge by clearly articulating results and ideas to customers, managers, and key decision makers.· Collaborate with product, engineering, and marketing teams across many Bosch business domains (automotive, industrial, internet of.

San Francisco, CA, US

• Created a tool to detect the location and type of golf swings using GPS and accelerometer sensor data as a consultant for the company Ped Motion.
• Engineered new features and trained several classifiers including Random Forests and Logistic regression models.
• My algorithm achieved an average distance error of 1.4 yards.
• Wrote about my development process in the following blog post.

Edmonton, Ab, CA

• Gained expertise in data analysis, data mining, machine learning, mathematical and statistical modeling, mathematical optimization and numerical methods.
• During my research training, I gained expertise in scientific computing using the programming languages R, Julia, Python, Maple & Matlab
• Among several other awards, I was the recipient of the prestigious NSERC CGSD3 Alexander Graham Bell Canada Graduate Scholarship. ($35,000/year for three years)

Edmonton, Ab, CA

• Teaching Assistant for the lab sections of Calculus 1, Calculus 2, Introduction toDifferential Equations and Linear Algebra.
• Completed the Graduate Teaching and Learning (GTL) program. As part of the GTL program, I assembled a teaching portfolio which includes my teaching philosophy, five reflections from over 100 hours of pedagogy sessions.
• Developed my presentation skills and my ability to discuss mathematical concepts in a language everyone can understand.

• Through a MITACS accelerate fellowship, I partnered with Maxfleet Solutions Ltd. to investigate the failure of drilling and fracking equipment.
• Used the relational database system MySQL to query over 10GB of sensor data.
• Implemented algorithms such as: independent component analysis, auto and cross-correlations and ARIMA models in both R and Python to detect anomalies in time-series data.
• Provided clear visualizations of key anomalies in time-series data using the ggplot2 and Matplotlib libraries.

Chiyoda-ku, Tokyo, JP

• I was selected to participate in the 2013 NII International Internship Program. I conducted research under the supervision of Dr. Ken Hayami and Dr. Akihiko Konagaya.
• Estimated parameters in pharmacokinetic models of the bile-duct cancer drug CPT-11.
• Used the Cluster-Newton method to solve underdetermined inverse problems.
• Collaborated closely with pharmacokineticists to discern key biological deviations in different types of cancer patients.
• Using a limited amount of data from 13 cancer patients, I generated over 1GB of simulated data to perform numerical experiments in Matlab.
• Extended the original Cluster-Newton method which lead to sped up in the convergence rate by a factor of 2.

Doctor Of Philosophy (Ph.D.), Computational And Applied Mathematics With A Certificate In Data Science

Bachelor’S Degree, Applied Mathematics, Minor In Physics With Distinction