We at Sber Robotics Center are creating a universal embodied AI for robots, which will control diverse physical embodiments: manipulators, mobile wheeled robots, robot dogs, mobile manipulators, and other robotic platforms.
We are currently looking for a middle/senior specialist to join our team.
The first stage of selection for this vacancy is a conversation with an AI recruiter. After applying, expect a message from them in SberChat; the conversation will take about 10 minutes. The AI recruiter's task is to clarify missing details and speed up the consideration of your candidacy. The AI recruiter is just starting out, so please be understanding. Your experience and participation will help make it convenient and useful for everyone!
Responsibilities
- Preparation and development of datasets for training VLM models in planning and robot interaction skills. Collection, cleaning, structuring, and validation of data from experiments, and model performance results.
- Formation of new training samples based on analysis of model errors and the need to expand system functionality.
- Development of pipelines for automating data collection, filtering, annotation, quality control, and versioning.
- Curation of data annotation: preparation of instructions, quality control.
- Interaction with ML researchers and engineers to prepare data for training, fine-tuning, and evaluation of models.
- Model training.
Requirements
- Experience with data preparation and processing for ML tasks.
- Proficient in Python and data manipulation libraries (Pandas, NumPy, and equivalents).
- Experience in automating data preparation processes and developing data processing pipelines.
- Understanding of dataset quality control principles and experiment reproducibility.
- Experience with Docker, Linux, Git.
- Understanding of machine learning and data preparation for model training.
- Skills in working with generative AI models; experience creating AI agents and using them in practice will be an advantage.
- Experience using GigaChat, Kandinsky, and equivalents in products.
- Instrumental command of AI for analysis, generation, and automation.
- Willingness to delve into the subject areas of robotics, VLM models, and multimodal data.
Will be a plus
- Experience preparing data for ML, NLP, CV, or multimodal models.
- Experience working with data for training LLM/VLM models.
- Experience with DVC, MLflow, ClearML, or other tools for managing experiments and datasets.
Conditions
- Office/hybrid format, address Avtozavodskaya 23a k2.
- Annual salary review and annual bonus.
- Extended voluntary medical insurance and preferential insurance for family members.
- Sber's unique training system for professional and career development.
- Favorable mortgage for employees.
- Prime subscription with the option to share with three close individuals.
- Referral bonus for recommending friends to the Sber team.
