Below I listed some of the projects I did in the past couple years. Each project either has Github page or webpage associated with the project
Review Microsoft Azure, ClarifAI and face_recognition
Why and What for? As a part of NLP lab at Emory, we were using face_recognition package based on dlib. However, the performance of the package on the dataset we had wasn't satisfactory enough. As a result, I was tasked to explore several packages online and determine which one performs better. After browsing online
What did I learn? This was the first time when I was tasked to do my own research and come up with analysis. Also, this was first time I implemented APIs and tested them.
Link to Github: Review of Face Recognition Packages
Front End Website
Why and What for? I designed small front-end website for local bakery in Kazakshtan.
Link to website: luckybakery.kz
When is the easiest time to land job?
Why and What for? Developed along with Rohan Bansal and Alvin Choi, this project was developed for DataFest Hackathon at Emory in 2018.
Using job applications data for job queries in the past 12 months, we came up with "Demand Index" in order to calculate how popular is the job in the given period. Then, using K-Means clustering algorithm we grouped together job queries in the particular region. The result, given new users input (job title and location) predict how popular is job in the over the past 12 months in the desired location.
What did I learn? This was the first project I have completed using both my statistical knowledge as well as CS. I also worked along in a team in the technical project under time constraint (48 hours)
Face Recognition for Webcam
Why and What for? During my sophomore year in college, I got interested in face recognition aspect of AI. I watched some online tutorials and videos. At first, I used Haar Cascade and Eigen Faces in order to recognize faces. The following techniques aren't state-of-the-art, but they are good for illustrative purposes.
What did I learn? I learned how to google and found some great resources, which I compiled in my github repository. Also now, I am using various other packages online that have accuracy rates of 95% and higher.
Link to Github: Face Recognition using OpenCV and Python
Emotion Detection using Text and Images
Why and What for? The aim of the project is to teach computers how to determine emotions based on texts and images. So far, most of the researchers have focused on emotion detection based on either text or images. This project is multi-modal because it focuses on combining both text and images.
What did I learn? I can't list all the things I learned in the lab, but some of the things are: face recognition and extraction, text and subtitle matching, video frame manipulation and conversion, and how to create HITs in the Amazon MTurk. Also, I got an opportunity to practice my coding skills in Python.
Link to Github: Emotion Detection
Machine Learning In Action
Why and What for? This pitch deck was designed for Emory Gouizeta Business School Pitch Competition. The premise of my pitch was to use Machine Learning (ML) and making it accessible for non-technical audience. In other words, create ML interactive dashboard for non-technical people to use and experiment with.
What did I learn? The project was big undertaking for 1 person - I was told my judges in the competition. However, this was my first experience pitching and trying to sell my idea to the investors.
Website for Orphanages in Kazakhstan
Why and What for? During my winter break of 2017, I wanted to built a website where various orphanages could request non-monetary products. More specifically, the website would have listings of various products such as books, medications, food and quantities demanded by orphanages. On the other side, people could have donated products and choose amount they want to donate. I contacted 10 different orphanages in the Almaty area and couple non-profit organizations. No one wanted to participate in the project, and people were afraid that due to corruption children wouldn't get products. By January 10th, before school started I had 4 different orphan houses that agreed to use my website. So far, no orphanages have contacted me to put requests on the website. Website is still running and is fully functional.