Senior AI Systems Engineer — Production LLM Pipelines

LandingAI, Inc.; United States; Remote; USD 150,000 - 210,000; Full time

Job summary

Haven Safety seeks engineers to own and optimize production LLM systems, designing end-to-end pipelines that blend model calls, tool usage, and retrieval. You will extend the agent orchestration layer to manage incidents from evidence through analysis and enterprise learning.

The role emphasizes evaluation discipline, context construction for LLMs, and production ownership from deployment to monitoring with strong Python and graph-based data skills.

Qualifications

• Production LLM systems experience shipping usable LLM systems.
• Hands-on experience building and debugging multi-step, tool-calling workflows.
• A repeatable approach to LLM evaluation with datasets and regression tests.
• Experience assembling context for LLMs and clear retrieval decisions.
• Proven ownership of systems from deployment to monitoring and incident response.
• Ability to discuss tradeoffs among quality, latency, cost, and context.

Responsibilities

• Production reasoning systems: build and operate multi-step LLM pipelines coordinating model calls, tool calls, and retrieval.
• Agent orchestration: extend Haven’s orchestration layer for incident handling from evidence to enterprise learning.
• Grounding and retrieval: design context layer across graph traversal and hybrid retrieval so models see the right evidence.
• Evaluation: build datasets, scoring, regression suites, and per-stage quality attribution.
• Production AI operations: implement tracing, cost and latency monitoring, and dashboards; catch drift and hallucinations.
• Technical direction: select models by task across major providers; partner with product and knowledge engineering.

Skills

Python programming

LLM orchestration

Debugging multi-step workflows

Tools

LangGraph

LangChain

Neo4j

Cypher

FastAPI

Job description

Haven Safety seeks engineers to own and optimize production LLM systems, designing end-to-end pipelines that blend model calls, tool usage, and retrieval. You will extend the agent orchestration layer to manage incidents from evidence through analysis and enterprise learning.

The role emphasizes evaluation discipline, context construction for LLMs, and production ownership from deployment to monitoring with strong Python and graph-based data skills.