James Irie

Experience

• 

  Marathon Oil Corporation

  Jan 2012 - Jan 2015

  Geophysicist

  Geophysicist in Upstream Technology group- Worked on data analyses and data pre-processing of large geophysical data. - Performed geophysical risk analysis for prospects. Reported findings to decision makers and business units.- Ran software assessments for the technology department.

• 

  JETRO USA

  Jul 2016 - Nov 2016

  Assistant of Business Development

  Created and managed a database of over 2500 US companies in 5 US states.Ran simple data analyses to find US companies that may have an interest in expanding to Asia.

• 

  Civitas Learning

  Dec 2017 - Dec 2023

  - Worked with product managers, software engineers, data engineers, dev-ops and data scientists to build new data science functionalities for civitas learning products.- Worked on building new machine learning infrastructures (ML pipelines / workflows) for existing and new prediction models.- Reconstructed ML codes from Spark to Metaflow- Developed ML models that outputs the probability of students graduating from their institution. - Developed ML models that outputs future term course enrollments. Tech: Python, Scikit-learn, Metaflow, Postgresql, MLflow, Pandas, Docker, AWS, Pytorch, Optuna Show less - Built an internal dashboard (AWS QuickSight) which monitored key model feature anomalies for multiple student subpopulations. Identified changes in feature values and data availabilities over time and their impact to our prediction models. Tech: AWS Step Function, AWS QuickSight, AWS Glue, AWS Athena, pandas, AWS Sagemaker- Built preliminary persistence prediction models for institutions who were transitioning to new learning management system (LMS). Analyzed and optimized models before productionizing.Tech: SQL, AWS Athena, Pandas, Matplotlibs, Jupyter Notebooks- Automated jupyter notebooks using papermill and multiple AWS. Set up Jupyter notebooks for automation processes and wrote AWS Lamda functions and AWS step Function workflows.Tech: papermill, AWS Athena, AWS Sagemaker, AWS Step Function, AWS Lamda, AWS S3, Jupyter notebooks- worked on custom impact analyses for student outreach programs which were conducted by academic institutions. Using propensity score matching technique with students' persistence prediction values, identified effectiveness of outreach campaigns.technology used:python, scikit-learn, pandas, jupyter notebooks, SQL, Amazon redshift,- developed internal data science frameworks and data science tools in python including feature anomaly detection, bias detection, clustering algorithm utilizing feature data availability and categorical feature engineering tools.technology used:python, scikit-learn, pandas, matplotlib, jupyter notebooks, Amazon S3, SQL, Amazon redshift- Ran EDA and preprocessing for internal data tools Show less

  • Machine Learning Engineer

    Jan 2021 - Dec 2023

  • Data Scientist

    Dec 2017 - Jan 2021
• 

  AffiniPay

  Mar 2024 - now

  Data Scientist