Serhii Shatov

This is a software system to serve neural networks and process tasks.

The engine uses message queues, a multitude of workers, and inference servers to process tasks in the swiftest way. The system is deployed inside Kubernetes, well-monitored, and can automatically handle traffic spikes by scaling workers and servers.

I was the data engineer, and my responsibilities were to write parts of it, insert new models, and maintain them.

The face-recognition software system can handle massive inputs and process them in real-time. It utilizes RabbitMQ as a messaging queue and has multiple workers to do the processing, including GPU-encoding workers and memory-heavy matching workers.

The system can accept updates in real-time, utilizing pre-processing device modules to work more efficiently. The system is used for multiple purposes, including brick-and-mortar stores, marketing analysis, and social networks.

I designed the system and wrote most parts, including the workers, databases, and the pre-processing device modules.

I developed the UAV-components communication for Kray technologies's UAV with nine motors, computer vision, and self-driving capabilities. I set up the ROS to communicate between components and worked on positional-data streaming and consuming.

I developed a few games before entering university and started working commercially. Some of them received awards at local contests. I designed 2D and 3D games, employing: ray-tracing technology, a voxel engine, AI simulation, and the A* navigation. I presented some technologies at the national paper defense and earned two bronze medals.