Omer Cerrahoglu

Integrated the TreeKEM cryptographic protocol into the company's instant messaging app. This cryptographic protocol supports groups of thousands of users, providing strong cryptographic guarantees, a la Signal's, while simultaneously providing scalable performance.

Worked on both the server and Android-based client. The server was written in Java and used Dropwizard over HTTPS and WebSockets to support a RESTful API, PostgreSQL for storage, and Redis for caching. The client-side was written in Java, and I added C++ code which I glued to the main codebase with JNI. I also worked on the front-end, improving UI performance for large groups.

Observed the correctness and performance. The work in correctness has been to ensure that the group remains consistent across the potentially thousands of devices with high performance. I implemented a performant message ordering system, ensuring clients receive messages both fast and in-order, and a group management system atop the cryptographic protocol, ensuring consistent group membership. For performance, my work has been developing new storage and caching algorithms, such as a logarithmic time algorithm to serialize and de-serialize cryptographic objects.

A full-stack solution for real-time live stream annotations. I was mainly responsible for the back end, including writing the WebSocket networking code, ensuring at-least-once semantics for Protobuf-encoded message transmission, optimizing video processing code to meet real-time standards, including designing and implementing the video processing pipeline, and developing a GStreamer-based solution for real-time, live feed annotations. I was also partly responsible for developing the OpenCV-based video processing code to detect motion in a live stream and the Qt-based client UI, including the code that would paste the above-mentioned motion annotations onto the live stream.

An administration panel solution for an on-site instant messaging application. The solution is meant to provide on-site administrators with useful tools to manage their application usage, including statistics on delivered messages, bandwidth usage, throughput, and clients' behavior. I worked on the back-end and wrote it in Java using Spring and Hibernate. I added various Rest APIs and written database queries to fetch statistics from the messaging server's databases.

A terminal-based chat client and server. I designed and implemented the WebSocket-based networking code and JSON-encoded message processing. The application has full support for all the basic functionality of a messaging application, including persistence (done through MongoDB), contact discovery, and support for multiple chats.