Human Rights on a Map
This month I and a group of very driven, talented and intelligent people worked on a web application project for a nonprofit organization. This organization graciously assists people who’s basic God given rights were not respected by the U.S government and private companies across the country. Human Rights First serve an important role to a lot of individuals who feel they have been treated wrong. Our team was tasked with creating a visual that highlights instances of excessive force used by police classified by a machine learning model served through our back end API being fed live data by Reddit.
TECHNICAL CHALLENGES
Scope of the project and tasked outlined for us we straight forward for the most in my case. The problem we are trying to solve was pretty clear. We want a visual feature of instances of police brutality as it’s reported and by user input on the web application. Implementing these ideas into code did prove rather difficult in practice however. I was individually tasked with creating a Fast API instance to pull data from Reddit, converting that data from a CSV into JSON format to be used by our talented web back end engineers to visualize that data. To highlight a few technical hurdles I encountered was actually building upon and sifting through code that was not written by me personally. It proved rather challenging navigating my way through code while implementing my own code and to not break any previous working code that was also being used. On the other hand I appreciate that challenge as to that provided experience that will aid me in future projects I work on here on out. Another technical hurdle I overcame was implementing problem solving features to make the data more readable for my teammates. I did a lot of cleaning and preparation of the data before being sent to web to be used. This data needed to be cleaned and condensed as to not take up too much space in our AWS deployed database. Optimizing my code and using various data science libraries such as Pandas and NumPY and specifically using Pandas to convert the data to usable JSON format was one of many techniques I used during the creation of the back end stream.
Another Set of Challenges Arise
A particular set of hurdles I went went through were more personal to me than anything. In the circle of computer science I hear a lot about this term called imposter syndrome, which is the feeling of doubting your skills, knowledge, or accomplishments. I believe this to be true in my case. I felt a lot of imposter syndrome during the progression of this project. I did not believe that I was able to accomplish the set of tasks I was assigned to do due to the fact that I was unsure if I would do it right. Implementing changes and making a proper ETL pipeline to feed data in a usable format for my team members was something of an impossible task in my mind. For anyone feeling that way just know you can overcome it by constantly reminding yourself that you are skilled enough to tackle any sort of project you do. That was a major battle that I am glad I conquered this project.
TEAMWORK
An important piece of completing a technical project that requires different skills and the meshing of those skills is teamwork. I was lucky enough to have such a wonderful team of fellow data scientist and web developers both front end and back end that were driven, intelligent and very understanding and willing to over communicate to actually understand what needs to be done. Communicating in these type of projects is paramount and I feel that this team was perfect in that regard.
WHAT I FORESEE IN THE FUTURE
Features that I would like to see added to this web application would expanding the map from just the US to maybe a the whole world which is very ambitious and would certainly come with its own set of problems both technically and morally I suppose. A lot of data would have to be collected and prepared for this endeavor. Our machine learning model we implemented would certainly have to be tuned to account for a lot variations and nuances that size of a datasets that includes just one whole country. Also what would be considered excessive use of force in another country. Different countries have different set of standards when it comes to policing the populace. That is a question that would be hard to answer and implement into a classification model that only gives a True or False answer. Another technical problem would be how would persist data of this size. Data from just one additional country would be several if not thousands of terabytes large. That is a lot of space being used. All in all I see a lot potential additions to this application that can help highlight the basic human rights violations in this world.
LESSONS LEARNED
I had a very expansive experience during this build. I learned a lot about how this project closely replicates the real world experience on a day to day bases. I appreciate that process. Other lessons learned were working with previous code and navigating the challenges that come with that. Its all very similar to real world experience and I appreciate that. I have learned that I can create stories and demonstrate the confidence that is needed to do the task according to my peers. The experience I have gained through this endeavor has helped me gain clarity and confidence that I will be do this in the real world and that what I have learned and the skills I have gained are actually being put to use in something that I truly enjoy that is solving problems.