Job details:

• Full-time
• Remote possible
• Start date: asap
• Duration: 6 month + extensions
• Long-term project
• Client: telecommunication company

Scope:

• Design and implement end-to-end GenAI solutions using Python.
• Build agentic workflows using frameworks such as LangGraph, tool/function calling, MCP, and multi-agent orchestration.
• Develop robust RAG pipelines including ingestion, chunking, embeddings, vector search, re-ranking, and grounding validation.
• Implement memory strategies (short-term, long-term, summarised, vector-based memory).
• Integrate and evaluate multiple LLM models (e.g., GPT-4.x series, GPT-4o).
• Develop structured output and tool-calling systems with schema enforcement.
• Build evaluation frameworks covering hallucination detection, quality scoring, latency, and cost metrics.
• Implement Responsible AI guardrails including input/output filtering and fallback strategies.
• Collaborate with DevOps/LLMOps teams for deployment, monitoring, scaling, and cost optimisation.

Skills:

• Strong hands-on Python development (mandatory).
• Experience with GenAI frameworks (LangChain, LangGraph, CrewAI).
•  Deep understanding of agentic reasoning, orchestration, and tool integration.
• Experience with Azure AI Foundry and Azure OpenAI.
• Familiarity with vector databases (Azure AI Search, Pinecone, Weaviate, OpenSearch).
• Understanding of tokenisation, context windows, rate limits, and cost optimisation.
• API integration, async programming, and resilient error handling patterns.
• Strong knowledge of Responsible AI risks and mitigation.

Other Requirements:

• Production-first mindset (observability, resilience, scalability).
• Ability to balance innovation with compliance and safety.
• Strong documentation and architectural communication skills.

SNI sp. z o.o. will process personal data for the purpose of the recruitment process in accordance with Data Privacy Policy. The data may also be stored and processed for future recruitment purposes, in accordance with the given consent.