Voice Agent AI Engineer

Be in IT is a recruitment company specializing in IT specialists. The recruiter has been active in IT recruitment since 2016 and has been running their own recruitment company since 2021, focusing on comprehensive support for clients' IT technology needs. The recruiter holds a certificate from the National Register of Employment Agencies (number 27663).

• Minimum 4 years of experience in Python in a production environment.
• Very good knowledge of asynchronous programming (asyncio, race conditions, concurrency).
• Experience working with LLM (function calling, prompt engineering, error analysis).
• Ability to design systems ensuring AI reliability (guardrails, validation, flow control).
• Practical knowledge of Pydantic v2.
• Experience testing nondeterministic systems (scenario tests, quality metrics).
• Experience with AI tools (e.g., Claude Code, Cursor, Copilot).
• Independence and ownership of solutions from design to deployment.

Nice to have:

• Experience with LangGraph or similar orchestration tools.
• Experience with Voice AI (e.g., Deepgram, Whisper, ElevenLabs).
• Knowledge of MCP (Model Context Protocol).
• Experience with Datadog, LangSmith, OpenTelemetry.
• Experience with DeepEval or other evaluation frameworks.
• Experience with Docker, Kubernetes, AWS.
• Experience working in a startup environment.

• Designing and developing AI agent workflows based on LangGraph (StateGraph).
• Building and optimizing transition logic, states, and decision paths.
• Implementing guardrails for LLM (detecting hallucinations, errors, missing sequences).
• Creating tools invoked by LLM using async Python, Pydantic, retry logic, and error handling.
• Designing and maintaining prompts and YAML templates with dynamic context.
• Developing and maintaining an evaluation framework (scenario tests, quality metrics).
• Integrating systems via MCP (HTTP/SSE, LiveKit RPC).
• Monitoring and optimizing performance (Datadog, LangSmith, latency <500 ms p95).
• Collaborating on voice pipeline components (STT, TTS, VAD, echo detection).
• Debugging and improving agent behavior in production.