— Have advanced knowledge in relevant areas of mathematics and statistics;
— Have a working knowledge of various techniques, tricks, and best practices for designing deep
nets, hyperparameter optimization, visualization, interpretation, etc;
— Have a bunch of failure stories about the deep learning projects you’ve been involved in (and
some success stories);
— Have a good understanding of deep generative models. You may work on autoregressive
models, GANs, VAEs, or flow-based models;

Will be a plus

— Knowledge of Digital Signal Processing theory;
— Knowledge of speech processing;
— Knowledge of CUDA / interest in developing custom CUDA kernels;
— Experience deploying machine learning models;
— DevOps experience (Kubernetes, etc.);

We offer

— Do real research with deep generative models of audio in a small, scientifically-minded team;
— Freedom to try out ideas of your own;
— There might be publication opportunities;
— Office in the center;


— Implement and maintain deep learning models in PyTorch;
— Train models, keep track of performance, and tune hyperparameters;
— Proactively read relevant literature and generate ideas for model enhancements;
— Help with deployment and ops as needed;


    Location: Kyiv
    Type: Full-time