Michael Thomas

• Integrated LLM functionality into analytics stack via Langchain and Llama Index, and internally designed web components.
• As founding Data Engineer, I built and scaled our data infrastructure to store, monitor and analyze terabytes of data
• Contributed to our web platform (Elixir)
• Contributed and helped design internal bidder pacing algorithm.

• Lead migration effort from in house ETL systems to Apache Airflow Architecture
• Developed and maintained internal APIs surfacing predictions and estimates from Data Science models to core product UI.
• Re-factored pandas code to production grade spark
• Developed CI systems for Data Science oriented work
• Developed unit and integration testing frameworks for data science workflows.

* Lead Fanduel's first Machine Learning Engineering team focused on building services bridging the gap between Data Science and Software Engineering.

* Architected internal feature-store (feast)

* Serve as mentor and consult for fellow engineers and management respectively.

• Pioneered the utilization of Spark's structured streaming (Databricks, Kinesis, Lambda, SNS,SQS).
• Develop batch ETL pipelines pulling from third party APIs and internal operational stores.
• Maintain Apache Airflow and Redshift AWS infrastructure.
• Deployed Fanduel's E2 Databricks Workspaces (Terraform)
• Deployed ML Models behind REST API (Fast API, ECS, RDS).

• Built and maintained Data Lake for global team (BigQuery/GCS).
• Published internal python packages for data scientist to leverage for pulling data from external APIs (Pandas/Python).
• Designed ETL pipelines for various end consumers (Google Cloud Composers/ Apache Airflow).
• Documented best practices for running queries against serverless architecture.
• Developed internal web application to develop personalized, re-usable, templates for panel study questionnaires (jQuerry, Flask, CloudSQL).

• Optimized/Managed media buying via open RTB.
• Automated various reporting tasks for client facing roles (Python).
• Developed dashboards/alarms to monitor data streams from clients' sites to internal data systems (Dynamo DB, Splunk, JavaScript).
• Assisted data science teams in targeting algorithm optimization. (Light GBM)