Company:** Sberbank PJSC Location: Moscow, RU Employment Type: Internship

Internship in Sberbank's Graph DS team: developing solutions for analyzing connected data — frameworks for reasoning LLM agents and ontology generation, as well as applying graph neural networks in antifraud tasks.

Tasks:

• Integration of LLM and knowledge graph, semantic search in the knowledge graph, framework component design.
• Extraction of concepts and relationships from text, automatic ontology expansion, validation and contradiction resolution.
• Solving graph partitioning and sampling problems for memory optimization, studying antifraud research, applying algorithms for generating synthetic transaction graphs.

Requirements:

• Full-time student of a Russian university, 3rd-6th year of Bachelor's/Specialist degree or 1st-2nd year of Master's degree.
• Knowledge of graph theory, graph neural networks, geometric deep learning, knowledge graphs (RDF/RDFX, OWL).
• Proficiency in Python, NumPy, PyTorch, Git, Bash, Data Science, Deep Learning, Machine Learning.
• English at upper-intermediate level.
• Mathematical statistics.
• Residence in Moscow or cities of the Moscow Oblast (Balashikha, Vidnoye, etc.).

Conditions:

• Paid internship.
• Duration: 3 months.
• Schedule: 40 hours per week.
• Hybrid work format.
• Comfortable modern office.

Skills:

• Graph Neural Networks
• Geometric Deep Learning
• Knowledge Graphs
• RDF
• RDFX
• OWL
• Data Science
• Deep Learning
• Machine Learning
• Git
• Python
• NumPy
• PyTorch
• Bash