Description
We develop solutions for detecting fake and generative content, video, and audio deepfakes. We conduct Disruptive Research and then roll it out into production.
We are looking for a Middle+/Senior-level Data Scientist for Audio Data Processing who will improve current detection methods, develop and implement new ones, and create SoTA solutions.
Responsibilities
- support and development of the existing audio deepfake detection solution, monitoring model degradation and fine-tuning
- development of new architectures and training of DL classifiers for audio files, building model ensembles
- writing scripts for automating audio analysis and general characterization - channel noise, number of speakers, speaker gender, detection of emotional tone, and more
- analysis and interpretation of model predictions, development of methods for visualizing interpretable features of audio forgery
- expert support for juniors and DEs in collecting and preparing new datasets for training
- studying and implementing methods and ideas from SoTA papers on the topic.
Requirements
- 3+ years of experience in DS
- experience with Deep Learning in Audio and other domains, experience using modern neural network architectures in Audio/CV
- proficiency in various neural network training methods: Incremental learning, Self-Supervised Learning, etc.
- understanding of approaches for feature formation for audio data analysis: Fourier decomposition, Mel-spectrograms, Wavelet transform, etc.
- understanding of modern end-to-end approaches to audio feature formation
- fluent in PyTorch
- proficiency in classical ML methods.
- knowledge of probability theory, mathematical statistics, and mathematical analysis.
Will be a plus:
- experience in projects related to emotion recognition from speech, speaker diarization, speech recognition
- experience in scientific work, working in a research unit, and writing scientific papers.
Conditions
- comfortable modern office in Oruzheiny Business Center
- hybrid work format after the probationary period
- annual salary review and annual bonus
- corporate gym and relaxation areas
- more than 400 educational programs from SberUniversity for professional and career development
- extended VHI, preferential insurance for family, and corporate pension program
- flexible mortgage discount equal to 1/3 of the Central Bank's key rate
- free SberPrime+ subscription, discounts on products from partner companies
- referral bonus for recommending friends to the Sber team.