Intern Researcher for Sber's Corporate Investment Business team: evaluating and developing LLMs and AI agents for banking products, conducting experiments, and publishing results.
Responsibilities:
- Design architectures for LLM systems, AI agents, and multi-agent solutions
- Create benchmarks and eval pipelines: assessing tool use, planning, and memory skills of AI agents
- Conduct experiments with models and agent scenarios
- Transform research results into publications at the level of leading global ML/AI conferences and journals
Requirements:
- Full-time student (3rd-4th year of Bachelor's/Specialist program or 1st-2nd year of Master's program) at a technical university with a strong mathematical background
- Interest in LLM research, not just using ready-made solutions
- Python programming skills and experience with version control systems (Git)
- Initiative, ability to decompose open-ended research tasks, and readiness to seek non-standard solutions in a rapidly changing environment
- Knowledge of classical optimization algorithms and probability theory
- Experience with LLM evals, AI agents, conducting research
Conditions:
- Work format: office or hybrid, Saint Petersburg, Uralskaya street, building 1, Sber IT hub (10 min walk from Vasileostrovskaya/Sportivnaya metro stations)
- Internship duration: 3 months, full-time (35-40 hours per week)
- Access to an extensive catalog of SberUniversity educational programs for professional growth and career development
- Flexible schedule and opportunity to combine internship with studies
- Adaptation program, internal training
