AI Agent Logic Engineer

About the Role

Our client are developing a platform where AI agents are assembled, configured, and deployed by non-technical users — yet under the surface, these agents rely on deep, carefully engineered logic.

This role sits exactly at that intersection.

You will not be building traditional backends or APIs. Instead, you will design and implement the decision-making layer that defines how AI agents think, react, switch states, and interact with external systems during real conversations.

⚠️ Important clarification:
This is not a Python-first engineering position. While general programming knowledge matters, your daily work will revolve around a custom scripting environment, built on top of Jinja2, used to express complex logic, data transformations, and conversation control. If your primary interest is writing application code in standard frameworks, this role will likely feel uncomfortable. If you enjoy mastering unconventional tools to shape complex system behavior, it will feel rewarding.

What You’ll Be Responsible For

• Designing agent behavior logic
  Build advanced conditional flows, branching paths, and state-aware logic that governs how an AI agent behaves across multi-turn conversations.
• Managing conversation state
  Ensure agents correctly track context, user intent, and system state across interruptions, retries, and edge cases.
• Integrating external systems
  Define how agents exchange data with third-party services (e.g. CRMs, calendars, booking tools) by constructing structured payloads and interpreting responses.
• Creating reusable logic components
  Develop configurable modules that can be reused across many agents by non-technical users without breaking consistency or reliability.

• Troubleshooting and performance tuning
  Investigate unexpected agent behavior (loops, dead ends, incorrect decisions) and optimize logic execution for fast response times in voice-based scenarios.

   

What We’re Looking For

• Strong logical and algorithmic thinking — you are comfortable reasoning about conditions, branching, recursion, and state machines
• Deep experience with templating or embedded logic systems (Jinja2 or similar), including non-trivial data manipulation inside templates
• High confidence working with JSON and complex nested data structures
• Willingness to learn and specialize in a non-standard, proprietary scripting language and become an internal expert

• Ability to work fluently in Russian, with professional-level English

   

Nice to Have

• Experience in solution engineering, implementation roles, or technical consulting
• Background with configuration-driven or low-code systems (e.g. workflow builders, automation platforms)

• Familiarity with Python or other general-purpose languages (for reasoning, not for day-to-day coding)

   

Why This Role Is Unusual

You won’t be shipping endpoints or UI features.
You will be shaping how AI agents reason, decide, and act — in systems where small logic mistakes can break an entire conversation.

If you enjoy building order inside complex, dynamic behavior — this role was designed for you.