My Projects

Why

With the amount of misinformation I've seen on social media about fad diets and fear mongering around health topics, I wanted to make a sciene-based site that I could share with my friends and that people I trust could contribute to.'

How

This site is self-hosted off my server with a astro frontend and a headless CMS backend called payload with its own dedicated site for contributors. Its been fully dockerized for easy development with auto-deployment to my server when I push to main.

Why

One of my many interests is signals and I thought this would be a fun way to visualize publically available data from the radio spectrum. I also wanted to learn more about software defined radios and how to process the data they receive.

How

The site and decoder were avalible Github projects that I edited to make the site more personal. The SDR is a USB dongle with an antenna that I have set up in my house to receive the ADS-B signals that planes broadcast. The data is then processed and sent to the frontend.

Why

While there's not much transoformative about the core functionality, I wanted something that fit my basic needs and also put me in control of the data security. Other sites send uploads to a server for processing- I travel alot and reformatting a photo of your passport or other sensitive documents maybe shouldnt be on safephotoresizer.rus.

How

This is a next.js app and it mainly uses the browser's built in canvas API to do the image processing. The document editor is built similarly except it uses a open-source library for exporting to PDF.

Why

I find AI very interesting and wanted to explore the security implications of it. As AI becomes more prevalent in security systems, I think it's important to understand how it can be easily tricked and why its so important to have a human-in-the-loop.

How

The frontend of this site is just a simple next.js app that uses the webcam to capture an image and send it to the backend for processing. The backend uses the ArcFace deep learning model and was trained on a large dataset. The attack uses the Projected Gradient Descent method to create false certainty.

Why

In collaboration with Brown Univerity, my team in a software engineering course designed this project to address the need for more accessible resources on socially responsible computing..

How

Using a Next.js frontend and a prisma/PostgreSQL backend, we created a robust admin and user auth flow hosted on Vercel and Neon db. Collaboration was a big part of this project as seen from the github repo tree.

Why

The need for the development of the automous drone function came from the limiutions set in the competion guidelines. I was in charge of drone avionics working alongside a team of all Mechanical Engineers.

How

Ardupilot is a robust and open-source software for controlling drones. We used a pixhawk flight controller and custom-built the drone frame and avionics. The software was intergrated with the hardware and tested extensively to ensure reliable flight.