ML/AI Engineer

Exon Group is a leading developer and integrator of innovative IT solutions in the construction industry.

Pulse is a corporate Data/AI Platform for forming a unified digital environment for the company.

The platform collects data from internal and external sources, brings it to a unified structure, links it together, and makes it accessible to employees, internal services, and AI agents. Pulse integrates data on projects, tasks, documents, communications, employees, organizational structure, business processes, decisions, and work events.

We are currently looking for an ML/AI Engineer to join our team.

Responsibilities:

• Design and develop the AI Runtime for Pulse. Develop RAG pipelines for working with documents, messages, tasks, reference materials, and corporate knowledge. Configure LLM pipelines for searching, extracting, verifying, and structuring facts. Develop agent scenarios: information search, context analysis, response preparation, tool interaction. Integrate LLM components with backend services of the platform. Work with embeddings, vector search, hybrid search, and reranking. Develop prompts, system instructions, tool calling scenarios, and evaluation sets. Develop mechanisms to reduce hallucinations: links to sources, evidence, human review, completeness check of the response. Research and implement API models and local models. Participate in the design of the Agent Platform and tool catalog.

Requirements:

• Mandatory: Python; Git; Docker; PostgreSQL; libraries for working with LLM API; libraries for processing text and documents; vector database or vector search; debugging and testing tools.
• Desired: OpenAI API / Anthropic API / local LLMs; Qdrant / pgvector / Weaviate / Milvus; LangChain / LlamaIndex / Haystack or independent RAG implementations; PyTorch; Transformers; FastAPI; OCR tools; ASR / diarization tools; reranking models; MLflow or analogues; observability tools for LLM applications.

Note: Candidates must know/be able to: understand LLM application architecture; design RAG pipelines for corporate data; work with embeddings, chunking, reranking, hybrid search, and query rewriting; understand LLM limitations: hallucinations, incomplete context, conflicting sources, quality degradation; build verifiable answers based on sources; design evaluation scenarios to check the quality of AI responses; extract structured facts from unstructured data; integrate AI components into backend services; work with documents, correspondence, tables, and text exports; consider access rights, data sensitivity, and audit trails; write clean, maintainable Python code.

Conditions:

• Official employment with a "white" salary;
• We value our employees' rights, so we provide a full social package according to the Labor Code of the Russian Federation;
• Remote work format throughout the Russian Federation;
• Decent salary level, to be discussed after the interview;
• Open corporate culture where your ideas can become reality;
• And we also have an active corporate life!