Data Scientist (LLM / Agents)

Moscow

Analytics

What you will do:

• Improve the quality of current tools: apply current approaches to obtain the necessary context, add support for multidimensional data (images, tables), work with agent policies.
• Add new agent tools to increase the percentage of automatically resolved requests.
• Develop the evaluation system: collect datasets and cases, develop metrics, build dialogue simulations, add a verifiable evaluation environment.
• Analyze errors from production and user feedback.
• Explore new approaches in AI agent development: memory, harness environment.

We expect you to have:

• 3+ years of experience in industrial ML development
• Proficiency in Python and a commitment to writing reproducible and high-quality code
• Experience in developing AI agents using frameworks (LangGraph / OpenAI Agents SDK / Google ADK), understanding modern design patterns: structured output, memory, planning, agent loops.
• Experience in building AI agent evaluation systems: LLM-as-Judge, FactScore, RAGAS.
• Experience in developing RAG systems: retrieval, chunking, sparse / dense / hybrid search, reranking.
• Understanding of ML/DL and NLP fundamentals: quality metrics, embeddings, general structure of transformers (tokenization, attention).
• Knowledge of algorithms and data structures, experience with the stack (FastAPI, Docker, Redis, Postgres, Airflow).

We offer:

• Voluntary medical insurance with dental care, telemedicine, and a mental health support program;
• Life insurance;
• Monthly Lamoda promo codes: two at 25%, two at 15%, and 40% on your birthday;
• A program of privileges and discounts from BestBenefits;
• Up to 15 days of sick leave per year paid at 100% of the current salary;
• Free legal consultations;
• Additional days off for important events;
• Exciting events and a lot of informal communication.

Development for everyone:

• An experienced mentor for every newcomer;
• Training and participation in external conferences;
• We help develop personal brands, speak at conferences, and write articles;
• Performance Reviews twice a year for systematic development;
• We hold meetups and demo days for knowledge sharing;
• We immerse you in fashion and introduce you to the latest trends every season.