AR Camera

• Built C++ framework for AR camera. Converted and Deployed on IOS. • Deployed method to render 3D masks on a face via OpenGL. • Applied Face Recognition/Tracking, Selfie Segmentation, Face Beautification. • Perform camera undistortion, worked closely with AVFoundation. • Developed Qt-applications for internal needs in computer vision team. • Reimplement core computer vision algorithms on GPU (Metal,OpenGL ES).

Snapmole

• Applied smart feature-fusing for RGB+Depth inputs from IPhone camera. • Researched and reproduced sota approaches. Trained MobileVitV2 on custom dataset and wrapped it to run inference on IOS. Trained segmentation model for masking skin leasions. • Managed AWS instance with 8 GPU to run PyTorch Distributed Data-Parallel Training. • Worked on dataset collection and annotation. Managed annotation team of 10 dermatologist.

Nufa

• Diffusion models research and development for AI avatars generation. • Image2Image, Text2Image, Adapters, LoRAs training. • Develop post-processing algorithms for smooth body change inpainting. • Develop smart tattoo generator, based on SD Inpaint. • Conduct research on Virtual TryOn. • Work on dataset gathering and filtering. Annotation team management.