AI Agent Engineer

Position Overview

We are seeking a Senior AI Agent Engineer to design, build, and scale autonomous multi-agent systems. In this role, you will go far beyond basic prompt engineering. You will architect robust state machines, manage complex agentic loops, and build the benchmarking harnesses that guarantee agent reliability. You will be responsible for

creating self-correcting, multi-step systems that can autonomously reason, execute tasks, and self-heal without getting stuck in infinite cycles.

Key Responsibilities

- Build Loop Architectures: Implement robust, cyclic state machines using tools

like LangGraph or custom execution engines to drive multi-step autonomous

behavior.

- Implement Self-Correction: Build reflection and critique loops where supervisor

agents evaluate worker agent outputs to ensure high-quality, autonomous

execution.

- Develop Evaluation Harnesses: Design and maintain strict benchmarking

harnesses (e.g., using DeepEval, Ragas, or custom frameworks) to track agent

trajectories and minimize regression.

- Optimize State & Memory: Engineer hierarchical memory systems (short-term,

episodic, long-term) and state-compression logic to optimize context window

usage.

- Establish Convergence Controls: Implement strict guardrails, deterministic





fallback logic, and exit criteria to prevent infinite loops and erratic agent behavior.

Required Skills & Qualifications

Technical Expertise

- Agentic Frameworks: Solid experience with LangGraph, Temporal, CrewAI,

or custom graph-based orchestration frameworks.

- State Machine & Graph Theory: Solid understanding of Directed Acyclic

Graphs (DAGs), statecharts, and deterministic routing.

- Harness & Eval Engineering: Proven experience building automated testing

harnesses to evaluate LLM outputs and agent trajectory traces.

- Advanced Prompting & Tooling: Expertise in structuring complex tool-use

(function calling) pipelines and dynamic context assembly.

Experience

- Production Deployment: 2+ years of experience deploying production-grade,

autonomous AI agents or complex distributed workflows.

- System Design: Background in computer science, software engineering, or

backend systems where deterministic logic meets probabilistic AI models.

Nice-to-Haves

- Experience with runtime sandboxing and secure execution environments for

autonomous agents.

- Deep familiarity with Anthropics Claude API, prompt caching, and context

window optimization.

📌 Artificial Intelligence Engineer - Dehradun
🏢 Cynet Systems
📍 Dehradun