Рaycheck: 120-180к net for full-time

Format: part-time (20 hours/week), remote work, part-time job at MIPT

Detail:
At the Compression Group, we thrive on the breadth of AI. To do this, we write methods for compressing models in PyTorch SotA: we optimize the costs of AI infrastructure and prepare models for transfer to new chips / devices / etc. After us, the customer’s bones are reduced, and more accurate SotA models are spinning in the devices (for example, our developments helped to put them on FairSeq devices).

We cooperate with experienced ML-engineers who want to realize themselves part-time in research work: to cut research projects with us, share experience with the team and poke SotA models in the pipe for testing new ideas.

Tasks:

• You will provide the project with ideas, you will quickly check them, you will solve the key challenges of the project and ensure the growth of target quality metrics;
• You will log the results and artifacts of experiments, prepare conclusions based on them, and make your work transparent and predictable for the team.

Knowledge and skills:

• You know the main architectures of NN, the principles of optimizing DL models (Quantization-Aware Training, Post-Training Quantization, Knowledge Distillation, Pruning etc);
• You know how to apply neural network compression methods in practice and work with specialized libraries (for example, torch.quantization). You can implement compression models and benchmarks from PyTorch articles from scratch;
• You can improve SotA mesh compression methods to achieve the goals of the task or work in new constraints;
• You know about the scientific approach, you know how to generate and test hypotheses, to produce reproducible and reliable results;
• Advanced use of Python; Git and IDE are your daily tools; you do not see any obstacles in working with a remote server; you know how to please the team with code culture and master team development practices.

It will be an advantage if:

• You have experience writing and editing scientific publications;
• You know how to identify customer needs and present a solution in terms of customer value.

Next steps:

• Leave an application on the site (P.S. include a portfolio and an open project in your resume);
• We will check the resume and portfolio, prepare a personal task;
• We will check the task and conduct an interview with the team.