Our team is engaged in the independent assessment and control of the Bank's models and AI solutions — from classic models and NLP to modern LLM solutions and agents. We check correctness, challenge developer approaches, propose improvements, and develop our own alternative solutions. We are also developing our own product — AIVA, an LLM agent that automizes validation.
We are currently looking for a Data Scientist for the validation department with a focus on analyzing, building, and validating models and LLM solutions used in various business processes of the Bank.
Responsibilities:
- understand the structure of models from different business domains (from scoring to speech recognition and biometrics)
- check the correctness of models and challenge developer approaches.
- develop alternative ML solutions, research new methods and evaluation metrics
- automate periodic model quality checks (monitoring)
- conduct research on SOTA approaches for key bank processes
- participate in the development of AIVA — our LLM-based validator agent: test scenarios, evaluate agent response quality, test on real validations
Technology Stack:
Core Stack:
python, pandas, pytorch
Bonus Points:
- Spark
- solid experience with DL (lightning, pytorch-distributed)
- advanced experience with LLM/GenAI (langchain, smolagents, faiss, huggingface, vllm, etc.)
Infrastructure:
- our own sizable Hadoop cluster 🌭
- GPUs — up to 32 H100 for training or up to 10 A100 for runtime (depending on tasks)
Requirements:
- higher education in computer science, engineering, or physics/mathematics
- 1+ year of experience in ML (we are ready to consider strong candidates with less experience but with good pet projects or competition participation)
- deep understanding of ML theory, mathematical statistics, and probability theory
- experience with major data analysis libraries.
- basic knowledge of bash and git.
Will be a plus:
- experience in building industrial ML solutions
- ability to decompose complex tasks and a proactive approach.
Conditions:
- comfortable modern office near Kutuzovskaya metro station
- option to choose a convenient schedule — office / hybrid
- annual salary review and annual bonus
- corporate gym and recreation areas
- over 400 educational programs from SberUniversity for professional and career development
- extended voluntary medical insurance, preferential family insurance, and a corporate pension program
- flexible mortgage discount equal to 1/3 of the Central Bank's key rate
- Prime subscription with the ability to share with three close contacts
- discounts on products from partner companies
- referral bonus for recommending friends to the Sber team.
