Andrew Steingass Email and Phone Number

• Developed a Python NLP algorithm to match equipment keys across datasets, reducing manual work from months to under a week. • Built machine learning models with XGBoost, LSTM, and regression for budget forecasting, enhancing accuracy and efficiency. • Created a data analytics application in TypeScript for exploratory analysis and statistical insights, saving thousands of hours. • Designed Python ETL processes for a data lake, supporting rapid visualization and enabling accurate reporting with Microsoft SQL views. • Leveraged Azure cloud products for scalable development and deployment.

Spearhead the Data team at My Footpath

• Developed an autonomous Python ETL pipeline featuring unit testing, comprehensive error reporting, and a manual data intake process, feeding into GCP PostgreSQL• Engineered a relational PostgreSQL database using a structured, iterative design approach and successfully deployed it on Google Cloud Platform with a local/server persistent proxy authentication system• Leveraged CI/CD processes to engineer a cost-efficient Flask e-signature application and deployed a docker on Google Cloud Run. This strategic approach significantly reduced the tool's cost from $1,000 per user to a few cents.• Integrated various REST APIs to extract and/or load client data, including Box file cloud storage, close.io CRM, Google Sheets, and Gmail, into our Python-based data infrastructure.• Conducted extensive data conversions, transitioning unstructured data into standardized SQL formatsfor seamless integration into our Big Data platform. • Developed the department's automation architecture from scratch while also enhancing management flexibility with interface-to-code collaborative features.• Designed a repository structure that is logical, modular, and scalable, incorporating environmental files for secure credential management. This configuration significantly streamlined the process of applying repetitive code solutions, reducing the time investment from hours to just minutes.• Implemented an agile project management system and analyzed its insights for process enhancement. This led to greatly improved clarity and communication between the CEO and our team.• Maintained security and data governance options including GCP permissions, credentials storage, and sensitive data access through the FERPA security framework.• Addressed long standing data integrity issues which enabled capture of previously lost clients; adding 7% leads• Managed and processed unclean datasets up to 60 GB in size of types xlsx, csv, json, xml, and pdf.

• Accessed Twitter API using R studio to clean and return over 20,000 tweets containing “online shopping”• Organized project goals, team communication, and inspired team members to go above project scope• Made concise proposal to obtain the only peer awarded grant for needed resources• Currently under review for publication by Institute for Global Business ResearchIn Summary:I downloaded tweets with the key words "online shopping" from the third weekend in January 2020 and January 2021 (before/during Covid). The goal was to identify Covid related trends and impact on online shopping public sentiment to then infer Northwest Indiana (NWI) business implications. I partnered with NWI Society of Innovators (http://www.nwisoi.org/home/) to sponsor my project and to whom I would report my findings to. -----------------------------------The tools used were:• RStudio to gather and clean the data of any characters like punctuation• A Latent Dirichlet Allocation (LDA) styled topic modeler to form a master list of words talked about using a stopword file to filter out unwated words like "I, and, the"• Linguistic Inquiry and Word Count (LIWC) to rate individual tweets in predefined categories such as "anger, i, positive emotion"• Advanced Excel original model using XlookupAfter downloading the tweets I deleted duplicate tweets and usernames to further clean and clean marketing skewed results. Once I had a conclusion, I wanted to validate my findings through manual analysis of tweets because I strongly believe a macro analysis alone is insufficient. A.K.a the word time has two meanings as seen in "delivery time" vs. "first time". With the Xlookup Excel model, I could type "food" to return tweets with food, and then furthermore "donate" to return any words that had both food and donate. It is important to note, while I was a strong leader that facilitated us to go above the expected project scope, this project could have not succeeded without my capable teammates.

• Used goals to achieve success in risk management techniques, and the broader vision• Created advanced excel spreadsheets to analyze correct item costing• Hands on accounting experience