> Markdown version of [/jobs/ext/3361861-ai-agent-developer](https://www.wearedevelopers.com/jobs/ext/3361861-ai-agent-developer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# AI Agent Developer

- **Company:** Itransition
- **Location:** La Unión, Spain (Remote available)
- **Contract:** Permanent contract
- **Skills:** Java (Programming Language), Artificial Intelligence, Automation of Tests, C Sharp (Programming Language), Video Game Development, Java Virtual Machine (JVM), Python (Programming Language), Large Language Models, Virtual Agents, Programming Languages
- **Published:** September 21, 2026
- **Apply:** https://www.indeed.com/viewjob?jk=0de71f327da4e294

## About the Role

* AI Application Engineering: Strong experience delivering LLM-powered features to production rather than to prototype, covering tool use, multi-turn agent loops, structured output, and management of context, latency and cost
 * Can ship agents as services with authentication, queues and persistent state
 * Evaluation & Quality: Expertise building evaluation sets and regression suites that detect when a prompt or model change has degraded results with a clear metrics proving that
 * Programming Languages: Proven expertise in Python
 * Code Context & Retrieval: Strong experience getting relevant source code in front of a model through repository indexing and symbol or structural search, with the possibility to recognise where embeddings are not the right instrument
 * Tooling & Integration: Experience with Model Context Protocol (MCP) or comparable tool-server design, including tool granularity and returning errors to the model in a form it can act on
 * Safety & Guardrails: Solid understanding of permission boundaries, dry-run modes and human review gates for agents that modify repositories and drive physical devices
 * Agent Observability: Familiarity with tracing agent runs, tracking token spend per feature, * Cross-language Reading: Ability to read Java and C# well enough to reason about a JVM automation engine and the Unity clients it drives
 * Test Automation Understanding: Solid understanding of what makes an automated test flaky (timing, ordering, state leakage, genuine product defects) and of why a passing test is not evidence that a defect was fixed
 * Domain & Industry Experience: Experience working within a game development environment or the mobile gaming industry, or with Unity

## Description

* Agent Architecture & Engineering: Develop and deploy production LLM-powered services in Python using multi-turn agent loops, structured outputs, and tool integration
 * Tooling & Protocol Integration: Design robust tool-server integrations (MCP or similar) connecting AI agents with the underlying Java core framework and device infrastructure
 * Code Context Retrieval: Build repository indexing and code search mechanisms to optimize context feeding for LLM models during test generation and execution
 * Evaluation, Safety & Observability: Establish continuous evaluation suites, model regression detection, safety guardrails, token tracking, and agent run tracing