ML Engineer at Alexander Thamm GmbH
Additional Description of Tasks
Created a data collection, data pre-processing, and data analysis pipeleine from scratch in order to predict the probability/likelihood of conversion using logistic regression, random forests, and XGBoost based on historic user-data with an emphasis on accurate yet explainable models within the retail market
Generated data science and data engineering teaching material (i.e. basics of data science, supervised/unsupervised methods, cloud computing, etc.) that my coworkers used to teach colleagues of one of the largest European phone companies
Created from scratch our company’s first ever product, which was a sales forecasting application, using Flask and Plotly Dash as our deployment and user interface, and used Python to implement algorithms such as Prophet (Bayesian forecasting method by Facebook), Convolution Neural Networks, and Recurrent Neural Networks
Developed and demoed our company's NLP-based product, which is an application that's used to analyze and correct improper spellings in the German language
Assisted in the creation of a pipeline process using Kubeflow and the implementation of Convolution Neural Networks to classify a patient as being healthy, positive for COVID-19, Pneumonia, and/or other diseases using their x-ray images
Researched and documented the benefits and drawbacks of 8 machine learning platforms (AWS Sagemaker, Azure Databricks, Kubeflow with MLflow, etc.) to create an easy-to-use scoring system based on exploration, pipelines, model management, deployment, batch inference jobs, and monitoring
Generated data science and data engineering teaching material (i.e. basics of data science, supervised/unsupervised methods, cloud computing, etc.) that my coworkers used to teach colleagues of one of the largest European phone companies
Created from scratch our company’s first ever product, which was a sales forecasting application, using Flask and Plotly Dash as our deployment and user interface, and used Python to implement algorithms such as Prophet (Bayesian forecasting method by Facebook), Convolution Neural Networks, and Recurrent Neural Networks
Developed and demoed our company's NLP-based product, which is an application that's used to analyze and correct improper spellings in the German language
Assisted in the creation of a pipeline process using Kubeflow and the implementation of Convolution Neural Networks to classify a patient as being healthy, positive for COVID-19, Pneumonia, and/or other diseases using their x-ray images
Researched and documented the benefits and drawbacks of 8 machine learning platforms (AWS Sagemaker, Azure Databricks, Kubeflow with MLflow, etc.) to create an easy-to-use scoring system based on exploration, pipelines, model management, deployment, batch inference jobs, and monitoring
- Location: Cologne, Germany
- Start Date: January 2020
- End Date: December 2020
- List of Relevant Technologies Used: Python, R, Linux OS, Git, AWS SageMaker, AWS CodeCommit, AWS S3, Docker, Kubernetes, Plotly Dash, Flask, Tensorflow, Keras, huggingface (NLP), XGBoost, Pandas, Numpy, ggplot2, dplyr