AI Platform Engineer: LLMs, Multi-Agent Systems

Primis; New York (NY); On-site; USD 100,000 - 150,000; Full time

Job summary

Primis is seeking an engineer in New York to build AI infrastructure and scalable frameworks for intelligent automation. You will create reusable SDKs and APIs, integrate LLM orchestration, and ensure observability in production systems.

The role requires solid Python skills and experience with agent systems, as well as collaboration with cross-functional teams. Additional responsibilities include embedding safety features and documenting engineering standards.

Qualifications

• Solid Python engineering skills with hands-on experience in AI systems.
• Experience with agent orchestration frameworks and multi-agent design.
• Proven ability to integrate LLMs into products.
• Platform engineering mindset to build usable SDKs and APIs.
• Experience with observability in AI systems.

Responsibilities

• Build AI infrastructure and scalable agent frameworks for intelligent automation.
• Create SDKs and APIs for engineering and business teams.
• Integrate LLM orchestration and routing into AI systems.
• Ensure full observability for agent workflows in production.
• Embed moderation and safety features from the start.
• Document system architecture and SDK usage.
• Collaborate with cross-functional teams to define best practices.

Skills

Python engineering skills

Agent orchestration frameworks

LLM integration

SDK and API development

Observability and tracing

AI safety concepts

Communication skills

Fintech exposure

Job description

Primis is seeking an engineer in New York to build AI infrastructure and scalable frameworks for intelligent automation. You will create reusable SDKs and APIs, integrate LLM orchestration, and ensure observability in production systems.

The role requires solid Python skills and experience with agent systems, as well as collaboration with cross-functional teams. Additional responsibilities include embedding safety features and documenting engineering standards.