Security Engineer

Syntic AI, Inc.; United States; Remote; USD 120,000 - 180,000; Full time

Job summary

Syntic AI, Inc. in the United States seeks a backend engineer to build systems that help AI agents plan work, call tools, use memory and recover when tasks fail. The role focuses on agent orchestration and making workflows reliable through observability and safety controls.

You will implement runtime support for planning, tool use, memory, and recovery, measure outcomes, and improve success rates. This position emphasizes safe interruptibility and handling failure modes when models interact with

Qualifications

• Production experience with LLM-powered applications or agents.
• Strong backend development ability in TypeScript or Python.
• Familiarity with failure modes in tool-using models.

Responsibilities

• Build systems that coordinate agent workflows and tool use.
• Implement runtime support for planning, memory and recovery.
• Measure task outcomes and use evaluation results to improve success rates.
• Make agent behavior visible and allow it to be interrupted safely.
• Address failure modes that arise when models use tools.

Skills

LLM-powered applications

TypeScript

Python

Failure modes

Job description

Role overview

Develop backend capabilities that help AI agents plan work, call tools, use memory and recover when tasks fail. The role centers on agent orchestration and improving reliability through measurable evaluations, observability and controls for safe behavior.

Responsibilities

• Build systems that coordinate agent workflows and tool use.
• Implement runtime support for planning, memory and recovery.
• Measure task outcomes and use evaluation results to improve success rates.
• Make agent behavior visible and allow it to be interrupted safely.
• Address failure modes that arise when models use tools.

Requirements

• Production experience with LLM-powered applications or agents.
• Strong backend development ability in TypeScript or Python.
• Familiarity with failure modes in tool-using models.