Description

We are GigaChat Alignment. We make the model useful and reliable: SFT/DPO, distillation into small models, LoRA service, metrics, and validation pipelines. We quickly test hypotheses, accelerate training, and roll out improvements to production — first for internal clients, then for all of Russia.

Directions

Improving SFT / DPO: testing new training approaches, accelerating pipelines, generating new data, distilling knowledge from large LLMs into small ones.

Developing GigaChat quality metrics, for example, by assessing the ability to solve international-level olympiad tasks. Developing internal LLM-AS-A-JUDGE.

Developing a LoRa training service for GigaChat and GigaEmbedder. Increasing the stability and reproducibility of runs, creating validation and data generation pipelines using LLMs.

For these roles, we are looking for a talented NLP Engineer to join us in improving and developing GigaChat. For all these experiments, we have a cluster with a large number of A/H 100s.

Responsibilities

• Managing and monitoring existing ML pipelines, ensuring their stable operation.
• Developing and supporting a role-based access service for users and services.
• Automating metric collection and model monitoring processes.
• Setting up and maintaining CI/CD processes for ML projects.
• Working with PostgreSQL for storing data, model metrics, and access information.
• Supporting and developing internal MLOps tools (similar to ClearML/MLflow).
• Setting up and supporting pipeline orchestration (Airflow or similar).

Requirements

• Excellent Python skills, deep understanding of FastAPI and service architecture.
• Experience with PostgreSQL and other DBMS.
• Knowledge of CI/CD processes and tools (GitLab CI, GitHub Actions, Jenkins, etc.).
• Skills in process automation, monitoring, and logging.
• Experience with MLOps frameworks (ClearML, MLflow, or similar) is a plus.
• Understanding and experience in setting up task orchestration (Airflow) is a plus.
• Frontend development experience or integration of metric visualizations and access management is a plus.

Conditions

• Remote work in Russia.
• Possibility of employment through an accredited IT company.
• Annual performance-based bonus of up to 6 salaries.
• Regular salary reviews.
• Corporate gym and relaxation areas.
• More than 400 SberUniversity programs for growth.
• Adaptation program and manager's assistance at the start.
• Largest DS&AI community — more than 600 of the bank's Data Scientists, regular knowledge exchange, experience sharing, and best practices, interactive lectures and masterclasses from leading universities and experts from technology companies, digest of the latest developments in DS&AI and reports from major world conferences, regular internal meetups.
• Extended voluntary health insurance, preferential insurance for family, corporate pension program.
• Mortgage for employees under a discount program.
• SberPrime+ and partner discounts.
• Referral bonus for recommending team members.