At Ozon, I build and operate services for the internal ML platform, including an artifact storage and versioning service and a proxy for NVIDIA Triton Inference Server. I also redesigned the artifact loading pipeline, cutting deployment time from about 1.5 hours to 20–30 minutes.
At GlowByte Consulting, I built Spring Boot backends for banking clients and contributed to an Apache Flink ETL product. Earlier, I worked on the Digitex Futures exchange at SmartDec and developed a Java/Spring incident-notification service during my internship at SberTech.

