• human and attribute detection using YOLO;• Deployed and tested open source models for different tasks (MMpose, OCR, Face Recognition);• semantic segmentation;• metric learning problem for image retrieval/matching application

• Analysis and taking of transcription metrics (WER);• Improvement of a pipeline for working with the language model (KenLM);• Analysis and selection the best method of noise suppression and VAD;• Writing Articles;• Implementation of a diarization system based on a resemblizer withan accuracy of 85%.