Something went wrong.

We've been notified of this error.

Need help? Check out our Help Centre.

In-Progress Projects

These are my current personal projects that I'm working on or have recently finished. This page will mainly list my machine learning and web development projects (leaving out app development and startup focused projects).

AnimeXP: Anime Recommendation via Machine Learning

Description: Via a combination of kaggle datasets and scraping multiple anime sites, I was able to accomplish an anime recommendation website. You're able to add an anime (or more if you'd like) to a list, then receive recommendations for those anime. Link: AnimeXP.io

My role: I worked with one other software engineer. He handled most of the NodeJS backend and NoSQL database, while I focused on the frontend and machine learning model. I created the frontend in React and Redux. The machine learning model was constantly iterated upon throughout the process. Version 0 of the machine learning model was a custom algorithm utilized Cosine Similarity as the similarity metric, and Jaccard Similarity as the weighting for the similarity metric. Version 1 utilized Singular Value Decomposition (Matrix Factorization). Version 2 is currently being worked on.

Main Technologies: React (Javascript), Redux (Javascript), NodeJS/Express (Javascript), MongoDB (NoSQL), and TensorFlow (Python), Keras (Python).


ShooterScout: Machine Learning-powered gunshot detection system

Description: ShooterScout uses machine learning to classifiy audio into two catergories: gunfire or nongunfire. Then based on the classification, the authorities' custom dashboard is notified.

My role: For the most part, I worked on the project independently. I designed and made the Android app. I created the web-based dashboard, and used CoreUI for designing it. I based the Machine Learning algorithm based on a Speech Recognition algorithm by the Arificial Intelligence research team, Ice9 Labs. Finally, the machine learning algorithms was hosted on my python/Flask server deployed on Heroku.

Main Technologies: React (Javascript), Redux (Javascript) Android (Java), TensorFlow (Python), Keras (Python), and Flask (Python).


The video on the right, shows a simple (and outdated) version of the project. The project has advanced since then. In future versions, I might try to implement vision-based gunshot detection (e.g. crowd running for their lives). Additionally, I might use an API to automatically call authorities and have a computer read the coordinates to the authority figure, but this would have to be for production purposes only.

Broaden.io: Platform for self-evaluation and tracking using online rubrics

Website Link

Description: Currently, collaborating with four other software engineers in order to create a Single Page Application/Website (SPA). The goal of this website is meant to allow individuals to learn any subject topic independently usingcrowdsourced rubrics. Individuals would upload a rubric that would detail how to become masterful of a certain area. Another individual would be able to use this rubric to see how much they have mastered the subject, and the steps they need to accomplishmastery.

My role: I'm design-lead and junior frontend developer on this team. For design we use Material UI. For frontend development, we use React and Redux.

Main Technologies: PostgresSQL, React/Redux, Express, and Node.


AnimeXP (OLD): Website for anime news, streaming, and manga reading.

Description: Due to the lack of an anime streaming website that is optimized for mobile users. I've decided to create an all around perfect website for mobile anime-streamers. The website works mostly-offline thanks to it being a progressive web app. This was originally partially made using MongoDB, Express, Handlebars, and Node.

Main Technologies: MongoDB, Express, React/Redux, and Node.

The screenshot is not of the current website, but instead the initial version that was created using MEHN stack.


AsIfSaid: Website for creating realistic quotes of anyone

Description: Specify a person that have been quoted a lot or have posted a lot online (e.g. a musical artist with a lot of songs). Then the website creates a quote/text that actually sounds like the specified person said it. This was originally partially made using Python and Flask. This uses the statistical idea of a Markov chain by using the probabilities of words co-occuring to determine which word to show next. It repeats this process until a sentence is created, then repeats sentence creation until a sufficient body of text is created.

Main Technologies: MongoDB, React/Redux, Express, and Node.

The screenshot is not of the website, but instead the python script in the terminal using Drake as an example.


Using Format