Contact Info

JavaScript, Tensorflow, React

Object Detection Application

This was my first project related to object detection. It had me delve into the sub field of Artificial Intelligence known as computer vision. Computer vision is a field which provides an avenue for computers to obtain and analyze useful information from sources such as static images, recorded video, and live video. These visual inputs are analyzed to provide both users and consumers with vital information in which they may need. The easiest way to picture computer vision is looking at AI as enabling computers to think while computer vision acts as an avenue for computers to see.

This project was built upon open source libraries which were provided by Tensorflow and ITP at NYU. The core libraries used were Coco SSD, ml5, and React. Coco SSD is a pre-trained object detection model provided by Tensorflow which allows a programmer to utilize the power of computer vision in their own applications. Ml5 is a library which provides additional pre trained models for a wide variety of applications which expand past CV. This provided an avenue for detecting objects within static images. React is a JavaScript library designed to allow developers to build unique user interfaces.

Knowledge of front end tools such as HTML and CSS were required for understanding the sourced templates used. JavaScript was the core language used in the actual implementation of the application. This was both a challenging and unique project due to me not having any academic experience with any of the used languages and topics. I do want to give credit to the Youtuber Nicholas Renotte who is responsible for sharing the project template as well as providing guidance throughout the process.

IoT, Security, Survey

Data Privacy Concerns in IoT Devices

IoT devices are more prevelant now than every before. Having access to the world wide web within seconds has allowed our society to advance rapidly over the last few decades. Unfortunately these devices do not come without a cost. Private companies who manufacture these devices are able to collect personal information through their buyable devices. Think about how much personal information is shared just through your phone. Your home address, where you work, how often you call family are just a few of the values that are accessible. This paper delves into the security practices of companies and aims to fully understand the opinions of college students when it comes to their use of IoT devices and any privacy conerns they may have.

This was the first semester long research project I participated in my graduate school studies. It was an amazing opportunity as our team consisted of older graduate students and was advised by Dr. Mohammad. We took a survey based approach and then analyzed our final data set of 60 participants using Python and libraries such as NumPy and Pandas. Mathematical knowledge was also crucial to our success. We utilized various data analysis techniques such as Chi-Square tests.

Upon the completion of our research, we arrived at a conclusion. As the number of devices connected to the Internet grows, so does the amount of data obtained by such devices. IoT researchers aim to build systems that talk to each other and share data freely to improve functionality. Although the conveniences provided are numerous, we must be cautious about our data privacy. Through our survey we successfully obtained a general understanding of data privacy concerns of the public. We found out that although a lot of IoT device owners consider IoT devices to improve functionality and provide convenience, they were equally concerned about data privacy as people who did not own IoT devices. We also found that a lot of the respondents of our survey indicated that they would be comfortable obtaining IoT devices from big tech companies. Although such companies can be trusted to an extent, we feel like this can create a monopoly and do more harm than good in the near future.

Note: If you are interested in reading our final paper, please reach out to me!

Python, PushShift API, Data Pre-processing

Keyword Data Analysis Tool (AKA Reddit Shaming Karen's Tool)

The Reddit Shaming Karen's Tool is designed to allow the user to discover potentially hidden data trends with a given data set in relation to a specified keyword. Our program is given a specified data set which is obtained using the Reddit Pushshift API [2], parses and passes the data to our data analysis script, and finally outputs the requested values onto a series of graphs. These values may be anything which relates to the Reddit post which had contained the previously given keyword. This then will act as a simplified connector between keyword usage trends and the user.

Our client, Dr. Florian Zach, had us design our program to suit his research needs which had the following restrictions set in place. The provided keyword, as our project title suggests, was to be set as Karen throughout the entire process. Additionally, our sourced subreddit data was only to be obtained from three groups which were as follows. Group 1: Delivery services, Group 2: Food employees in general and firm specific, and Group 3: Restaurants. Thirdly, our program was to only look at data within a set time period during the start of the Covid-10 Pandemic. Lastly, we were required to implement some form of topic modeling to group the posts and comments appropriately. These restrictions were given to allow our client to understand how the pandemic affected the usage of Karen over a time period spanning many months.

The main goal of this project was to gain an initial understanding of online shaming behavior and also discover potential trends in the usages of the given keyword. As mentioned earlier, Karen will be the associated keyword, although the project will support Dr. Zach with measuring data trends with any keywords he may require. Reddit will act as our sole source of information. Reddit is a social media platform or forum, where unlike other platforms such as Facebook, users gather in given communities (known as subreddits), that discuss whatever the topic of that community is. For example, there are subreddits dedicated to discussing cats, music, and lots of other popular topics [6]. Our initial design will analyze a set size of subreddits which will be explained in greater detail in the Design section. In technical terms, the primary task of this project was to create a Python program that scrapes both current and historical Reddit data for specific keyword analysis using the PushShift API[3]. This would provide an avenue to understanding both current and past online shaming behavior. This API is designed to provide programmers with enhanced functionality and search capabilities when analyzing Reddit post data. The scraped data is then converted into the CSV and JSON formats, and parsed for all relevant information. This parsed information is used to create various graphs and charts which show keyword use trends over a given period of time. For this specific project, we will be analyzing the use of the keyword Karen 18 months before and after the start of the pandemic in March 2020.

Technical Interview, Hiring Practices, Whiteboarding

Faults of Whiteboarding: How Can SE Hiring Practices be Improved?

I am currently revising this projects description. Sorry for the inconvenience.

Java, Sourced Data Set

Covid-19 Data Manager

Currently waiting for the publication approval! Will update soon