Anthony Castro-Yoskovich

Analytics practitioner with experience solving problems using data and visualizing quantitative information.

Experience

Analyzed unstructured data to support analysis of a document dump containing thousands of PDFs for a legal client.
- Saved weeks worth of manual work by creating an excel deliverable that could be searched in familiar format.
- Used spacy and regex to identify NLP named entities and other custom text patterns (ex. search warrant IDs).
- Classified documents into various types using AWS Textract and enabled reader to sort documents chronologically.
Automated report creation to increase throughput 50x (4 / month -> 200 / month) for an internal valuation team.
- Built a web application using Python to support a human review process.
- Ingested data from emails (yuck!) and automated report creation using jinja2.
- Built diagnostic plots with plotly to identify outliers.
Led development of a Python package enabling the creation of Tableau Extensions using Python instead of JS.
- Built and deployed documentation with Sphinx and GitHub Pages (github.com/trexjacket).
- Worked with dashboard developers to create Tableau Extensions that enabled interactive data analytics.
- Worked collaboratively using git and GitHub, managed projects using Issues and Pull Requests.

Developed a risk-adjusted linear model in Python to classify more than 50,000 attendance rates.
- Visualized results in Tableau and published to company instance of Tableau Server.
Consulted with SMEs to leverage data to prioritize and inform classroom instruction.
Automated the conversion of over 1,000 PDF files into a tabular format using the Tesseract OCR engine in R.
- Saved more than 80 hours of manual work each year.
Consistently analyzed and visualized complex testing results of 4 subjects at more than 50 schools using ggplot2.

Delivered data-driven hiring and relocation insights by modeling $60 million worth of incremental sales.
Used Python, R, Tableau, and SAS to clean, merge, analyze, and summarize over 50 million rows of data.
Worked on a cross-functional team to deploy a floor space optimization tool (PuLP) and recommender system.

Occasional blogging and tinkering around on my website built with Hugo and hosted on AWS Amplify. View
Scraped all historic Amazon shareholder letters in Python, used natural language processing to analyze text, and visualized results with an interactive dashboard using Dash. View
Developed and deployed an interactive cost saving dashboard using Python, Heroku, and Bokeh. View
Created a Python script to log my internet speed and analyzed 800,000 data-points with pandas and seaborn. View
Built a Raspberry Pi data logger with Python to archive the price of Bitcoin using the Cryptowatch API. View

^*This document was created entirely using R Markdown.