Matthew Phillips Email & Phone Number

San Francisco, California, Us

• Using change data capture events from a database, set up streaming data pipeline to our feature store to reduce system latency.

San Francisco, California, Us

• Closely partner with data science to architect and lead development of customer-facing ML services and the ML platform, which includes a feature store and related backend services.• Architect and lead revamped MLOps pipeline on Vertex AI for key ML models in production, increasing reliability and reducing time to value from days to hours.• Hire, manage, and mentor a team of three MLOps engineers of different levels.• Own $700k/year data and ML platform budget, taking cost cutting measures while maintaining developer productivity. Lead company-wide migration to a new data warehouse and cost optimization efforts resulting in savings of $400k in the first two years.• Onboard dbt and decentralize the creation and maintenance of production analytics tables.• Improve data infrastructure by introducing data quality checks and better monitoring and alerting of data pipelines.

San Francisco, Ca, Us

• Establish the data engineering team; hire, lead, mentor, and manage a team of six MLOps and data engineers.• Bootstrap cross-functional MLOps team to improve data scientist productivity and robustness of ML model deployments.• Lead initiative to replace a third-party data science platform with an in-house Kubernetes-based platform.• Overhaul existing reporting data stack to use a modern data stack of Fivetran, Airflow, Snowflake, and Looker. Migrated reporting to Snowflake and achieved upwards of 200% increase in dashboard query performance.• Engineered new method of measuring impact of product features on key metrics by parsing application logs, which became the authoritative source for Operations and Product reporting.

San Francisco, California, Us

• Built data science platform on AWS to enhance productivity of dozens of data scientists using infrastructure as code tools CloudFormation and Terraform.• Deployed batch and realtime machine learning models with Docker, Flask, and AWS SageMaker, including SoFi's first customer-facing ML model: a credit risk scoring model which increased quality loan applicant volume by an estimated 1%.• Leveraged model interpretability techniques for machine learning in credit risk scoring.• Built data pipelines on TBs of data to populate enterprise data lake using AWS Glue and Athena.• Maintained enterprise-wide Kafka infrastructure.

San Francisco, California, Us

As SoFi's second data scientist, I was both a hands-on machine learning engineer and modeler on various projects across marketing, operations, and credit teams.• Developed machine learning model to improve prospect targeting, beating industry metrics by 10%.• Developed and delivered training sessions on various topics in machine learning concepts (deep learning) and tools (Spark, Keras) to the analytics team.• Reimplemented pricing algorithm in PySpark for individualized messaging to prospects.• Built predictive targeting models using Spark ML on TBs of data to support marketing initiatives.

New York, Ny, Us

• Lead instructor for Part-time Data Science Immersive.

Santa Clara, Ca, Us

• Modeled topics and classified tens of thousands of documents of conversational text using latent Dirichlet allocation and random forest in scikit-learn.• Built a credit scoring model for early stage venture lending using regularized GLMs.• Worked with key stakeholders to identify data science projects and grow the data science internship program to more than 5 interns.

West Haven, Ct, Us

• Topics included maximum likelihood estimation and computational methods.

Arlington, Virginia, Us

Education record

Master Of Science (M.S.), Data Science

Master Of Arts (M.A.), Economics

Bachelor Of Arts (B.A.), Political Science