Travis Bender

Background

About Travis Bender

I am currently working at CPKC as a Data Scientist. I work across our on premise and cloud environments developing and implementing data pipelines & machine learning solutions. I am experienced with AWS (S3, Sagemaker, Athena, cloud formation, RDS, permissions management, lambda, etc), python, spark/pyspark, pytorch, sqlalchemy core/orm, various sql dialects (Postgres, Sql Server, Oracle, Athena, Spark Sql, DuckDB, SAP Hana), database query/table optimization, computer vision, qliksense/view GIS/spatial data, Linux, power BI, columnar data storage, and serverless architecture. I regularly use a variety of machine learning techniques in production environments, including tree based models (XGBoost, Catboost, classification trees, random forests, etc.), Neural Networks, linear regression (Lasso, elastic net, traditional multiple linear), logistic regression, explainability techniques (primarily Shapely Values/SHAP). I've designed various ETL workflows for relational databases, including creating the schemas, tables, indices, partitions, archival/deletion strategies, and complex insert/update logic. I've built extensive tooling to streamline my teams work across the varied environments we deal with on a day to day basis. I'm the maintainer and developer of several productionalized machine learning/ai tools at CPKC-Arrival Time Estimation -Computer Vision based defect detection -Near real time estimated revenue recognition -Train consist based incident likelihood I've also built several data pipelines -CSV to parquet consolidation leveraging partitioned table structures for large scale IoT data (originally with AWS Glue, later with lambda and Athena for cost savings)-Near real time geofencing, with capability to add new geofences and reprocess historical data across (several TB compressed) within minutes. I've prepared datasets, statistical reports, and gathered data to respond to adhoc regulatory requests throughout the CPKC merger process. I've also prepared reports used in final offer arbitration cases. I regularly mentor and educate people across the organization to improve access to IS resources and democratize data science at CPKC. The processes I've built have delivered significant financial benefit to the organization, recouping 5+ million dollars/year through various mechanisms that are expected to continue in perpetuity.