Responsibilities

• Integrate existing LLM solutions into company products and services (via API, backend, data processing pipelines).
• Evaluate model response quality using automatic metrics (BLEU, ROUGE, MMLU) and participate in organizing human evaluation.
• Ensure safety and filtering of generated content (setting up basic guardrails).
• Work with RAG systems:

- Assist in data indexing; - Execute queries to vector databases; - Participate in optimizing search and generation under the guidance of senior specialists.

• Interact with product and ML teams to discuss technical solutions.
• Optimize inference performance at a basic level (setting up batching, caching simple results).
• Work with ready-made multimodal solutions (text, images, documents) as needed.
• Perform fine-tuning of existing LLMs using established pipelines and methodologies.

Requirements

• Higher education (IT, technical, mathematical).
• Proficient in Python and SQL.
• Good knowledge in:

- Probability theory and mathematical statistics; - Basic machine learning algorithms (ML); - NLP algorithms (text classification, tokenization, vector representations).

• Understanding of LLM operation principles, basic usage scenarios, and fine-tuning.
• Experience deploying ML solutions (Docker, basic scenarios in Kubernetes or cloud platforms).
• Experience with:

- Vector databases (executing queries, understanding index structure); - LangChain or LlamaIndex frameworks (basic usage); - LLM evaluation methods (understanding BLEU, ROUGE and their limitations).