AI Voice & Speech Engineer

Syntic AI, Inc.; United States; Remote; USD 140,000 - 190,000; Full time

Job summary

Syntic AI, Inc. is seeking an engineer to design and implement runtimes for AI agents, focusing on orchestration, planning, memory, and recovery. The role emphasizes safety, observability, and robust failure handling in tool-using models.

The candidate will address performance evaluation, tool integration, and system reliability to ensure observable and interruptible agent behavior. Strong backend skills in TS or Python are essential.

Qualifications

• Experience building production applications or agents powered by large language models.
• Strong backend engineering skills in TypeScript or Python.
• Understanding of failure modes in tool-using models.

Responsibilities

• Design and implement agent orchestration and tool-calling systems.
• Build runtime capabilities for planning, memory, and recovery.
• Measure agent task success and improve performance using evaluations.
• Make agent behavior observable and interruptible, with safety in mind.
• Address failure modes in models that use tools.

Skills

Backend development

Agent orchestration

Tools

TypeScript

Python

Job description

Role overview

The available description focuses on building a runtime for AI agents that supports planning, tool use, memory, and recovery. The work spans orchestration, performance evaluation, and safe, observable behavior; the source does not specify speech or voice responsibilities.

Responsibilities

• Design and implement agent orchestration and tool-calling systems.
• Build runtime capabilities for planning, memory, and recovery.
• Measure agent task success and improve performance using evaluations.
• Make agent behavior observable and interruptible, with safety in mind.
• Address failure modes in models that use tools.

Requirements

• Experience building production applications or agents powered by large language models.
• Strong backend engineering skills in TypeScript or Python.
• Understanding of failure modes in tool-using models.