Senior AI Engineer

ID 127762
127763
AI Engineer
Agentic Systems
Senior
"Customer location: Romania
Candidates locations: EU, All except Belarus, Russia
Language: English B2
Estimated start date: 23.11.2026
Duration: 12 months"
Project Responsibilities

Design, build and run LLM agents in production, both client-facing and internal
Build MCP servers on top of existing databases and production systems, and orchestrate several internal and external MCP servers under one agent
Design memory, context management and RAG for production use
Set up evaluation, handle edge cases and improve agents step by step after release
Work in an architecture that combines databases, interfaces, agents and orchestrator applications
Work on Amazon Bedrock and/or Microsoft Foundry
Candidate's Portrait

A hands-on AI engineer, Middle+ to Senior level, who has already shipped agents to production and can explain how they fail and how they were fixed. Ambitious, learning fast, experimenting with new tools and models on their own time. Not a classic backend developer who recently added "AI" to the CV.
At least 1.5 to 2 years hands-on with LLMs and agents in production with background in ML/Python/DS
Must-haves

6+ years of overall experience in IT field as Software Engineer, ML Engineer, AI Engineer.
Hands-on production experience with Amazon Bedrock and/or Microsoft Foundry (Azure AI Foundry, Azure OpenAI)
Built and shipped LLM agents to production, both client-facing (customer assistants, conversational AI) and internal workflow automation agents
Strong experience with at least one agent framework: LangGraph, LangChain, LlamaIndex, Semantic Kernel, AutoGen, Strands Agents or OpenAI Agents SDK
Deep MCP experience: building MCP servers on top of existing databases and systems, integrating external MCP servers and orchestrating several of them under one agent
Production-grade RAG, memory and context engineering: embeddings, chunking, hybrid search, reranking, vector databases, short- and long-term agent memory
Can name and explain agentic patterns personally implemented (ReAct, planner-executor, reflection, router, supervisor, multi-agent, human-in-the-loop)
Experience with LLM evaluation, guardrails and monitoring of agents in production, and iterative improvement based on real usage and failure cases
Understanding of system architecture combining databases, APIs, agents and orchestrator applications, plus practical Docker and CI/CD experience
Nice-to-have

Trading, investing, brokerage or crypto domain knowledge combined with AI engineering. Top priority for the first batch: such candidates go first.
Fintech or regulated-industry background (compliance, data protection, auditability of AI outputs)
Broader AWS or Azure cloud experience around the Bedrock/Foundry stack
Experience with LLM evaluation and observability tools (e.g. LangSmith, Langfuse, Ragas, custom eval pipelines)
Deep practical mastery of agentic development tools used to deliver full prototypes quickly. Counts only on top of the must-haves, never as the main selling point.