Tech Stack: Fluent in Python, C++

Expert’s cases:

1. Joined the team at the very beginning of a startup, implemented many projects from scratch to production.

2. Developed ML pipelines for various tasks:

   • Object instance segmentation with F1 score from 0.87 to 0.96 on several benchmarks.

   • Object size assessment (via custom classification NN) with Accuracy of 0.71.

   • Custom combined model (YOLOv7) for both instance segmentation (with F1 of 0.85-0.95) & size assessment (with Accuracy of 0.85).

   • Semantic segmentation with F1 score from 0.84 to 0.94 on several benchmarks.

3. Contributed to design and implemented a method of comparing image quality and a real-time object tracking algorithm.

4. Contributed to model deployment on local hardware (successfully piloted in 2 diagnostic centers) and remote servers, first in Python, later in C++.

5. Analyzed model errors, maintained data labeling pipeline, communicated with medical professionals (labelers), and assigned labeling tasks.

6. Wrote technical description of ML part of product patent, which he co-authored.