Description

Our team creates control algorithms for Sber's own anthropomorphic robot. The current focus is on whole-body control, training a neural network controller that can execute any command - from waving hello to dancing and acrobatics.

Responsibilities

1. Create robot control algorithms based on training the robot model in a simulator using Reinforcement Learning and Imitation Learning methods
2. Implement and adapt SOTA papers from the field
3. Test and debug algorithms on a real robot
4. Improve the robot model and learning environment, solve the Sim2Real problem

Requirements

1. Strong development skills using PyTorch
2. Experience using Reinforcement Learning, Imitation Learning approaches
3. Experience working with MuJoCo, Isaac Sim simulators
4. Experience in robotics, with real robots, is desirable
5. A plus is having publications in ICRA/IROS/ICML/NeurIPS and other conferences and journals on Robotics and DL

Conditions

• Friendly and highly qualified team
• Unique large-scale projects, work in a priority direction
• Competitive salary (base salary + annual bonus)
• Modern workplaces and software
• VHI, corporate pension program, accident insurance, social guarantees, corporate events
• High level of corporate culture
• Work in the office (Moscow, Kutuzovskaya metro station), possibility of a hybrid schedule