Brian Seel

Baltimore City has a challenging relationship with its transit provider, requiring multiple workarounds to complete proper reporting. The city needs the data to understand how the system operates and for federally mandated reports. This project scrapes data from the provider's website, parses various spreadsheets with dirty data, and cleans it up for internal use in an agency database.

I also built a Python wheel for our data provider's undocumented API, which required a bit of reverse engineering to understand how it worked. The code is available here: https://github.com/city-of-baltimore/ridesystems.

The scraper: https://github.com/city-of-baltimore/ridesystems/blob/main/src/ridesystems/reports.py.

This project was less impressive from a technical perspective, but it made a difference in my city. Baltimore had a bike share system with chronic issues, and the operator was unwilling to admit that the problems were as big as they were. I wrote this script to regularly pull the number of available bikes in the city, which showed that the numbers were quickly falling to levels well below the service level agreement.

I used that data to write a published article highlighting the problems, which resulted in the system being shut down three weeks later.

Articles:
• https://baltimorefishbowl.com/stories/the-baltimore-bike-share-system-is-about-to-fail-again/
• https://www.baltimoresun.com/business/bs-md-bike-share-ends-20180815-story.html

After our local bus system opened its real-time tracking data to the public, a few friends and I wrote a system to track, store, and visualize the data that was becoming available. This project was necessary because the data API only shared real-time data and not historical data. Throughout two days, three of us used Elasticsearch, Logstash, Kibana, and AWS to track that data and better analyze the historical trends.

An additional write-up is available here: https://cylus.org/how-we-made-a-website-to-track-our-bus-system-in-36-hours-e9870c3c3a38.

Some examples of the data were published here: https://cylus.org/mtas-on-time-performance-data-for-april-a378c86bd666.

This project was part of a social engineering challenge where the goal was to exfil a file from the desktop of a target. The game had another challenge that involved drawing bubbles on the screen, so I delivered a Python script that was supposed to solve that problem while doing the file collection. The code connects to an IRC channel that exfils the data to a file called "flag.txt" on the user's desktop. It can be extended to accept commands and maintain persistence.

After Baltimore's Automated Traffic Violation Enforcement System was shut down in 2015, there was a higher level of scrutiny on the system when it was brought back in 2017. ATVES had data spread across many different providers, different vendors administered the red light and speed camera programs, and the financial data was stored separately.

This project aimed to bring the data together to be tracked on internal dashboards, which required three different scrapers and a data cleaning before being brought into our internal database. It required three scrapers to pull the data from our Red Light Camera vendor, our Speed Camera vendor, and our financial information. The financial information was behind a Windows-authenticated website, which required extensive reverse engineering and patching an external library to get it to work.

https://github.com/city-of-baltimore/atves/blob/main/src/atves/axsis.py
https://github.com/city-of-baltimore/atves/blob/main/src/atves/conduent.py
https://github.com/city-of-baltimore/atves/blob/main/src/atves/financial.py

One of our team's struggles was that we would develop dashboards, and some users wouldn't have permission to use them or forget to check them. The customer requested that we send screenshots of some of the dashboards, so I built this tool to take screenshots and email them to the relevant people automatically.

This project used Selenium to take the screenshots and the SMTP library to email the screenshots out.

Our vendor of parking citations had their data available to authenticated users but did not have a good way to provide the data on a rolling basis to us. This scraper pulls down the data and puts it in a database for later use. https://github.com/city-of-baltimore/ticketstat/blob/master/parkingcitations/gtechna.py