LLM Agent Orchestration Engineer

Xora Innovation; San Diego (CA); Hybrid; USD 140,000 - 200,000; Full time

Job summary

ELEMYNT, an Xora Innovation spinout, is building the foundational layer for LLM-powered features across our platform. You will design agents that plan, call tools, and reason over results, then make them dependable in production with traceability and guardrails.

In a fast-moving startup, you’ll implement provider abstractions, agent orchestration, retrieval, and evaluation, shipping reliable, auditable workflows across hybrid on‑prem and cloud environments.

Qualifications

• Bachelor's or Master's in CS; 5+ years shipping production software.
• Strong Python; async code, typing, testing, modular design.
• Hands-on with agent/LLM systems in production.
• Experience with multiple model providers behind a single abstraction.
• Experience end-to-end retrieval systems: embeddings, chunking, hybrid search.
• Observability: tracing, prompt versioning, debugging real behavior.
• Comfort owning ambiguous systems in a fast-moving startup.

Responsibilities

• Build provider abstraction for multiple models with validation, retries, cost tracking.
• Build agent orchestration with planning and sub-agents, checkpoints.
• Create human-in-the-loop checkpoints for high-stakes steps.
• Wrap platform capabilities as typed, registered tools.
• Design retrieval end-to-end: embeddings, chunking, hybrid search, grounding.
• Build versioned prompts and captured reasoning with traceability.
• Expose guarded model access as tools across backend/frontend/notebooks.
• Instrument model calls and tool invocations with tracing and lineage.
• Develop evaluation framework with LLM-as-judge scoring and gating.

Skills

Python

Async code

Typing

Testing

System design

LLM systems

Observability

Distributed systems

Education

Bachelor's or Master's in CS

Tools

LangGraph

AutoGen

Temporal

MLflow

Langfuse

Job description

ELEMYNT, an Xora Innovation spinout, is building the foundational layer for LLM-powered features across our platform. You will design agents that plan, call tools, and reason over results, then make them dependable in production with traceability and guardrails.

In a fast-moving startup, you’ll implement provider abstractions, agent orchestration, retrieval, and evaluation, shipping reliable, auditable workflows across hybrid on‑prem and cloud environments.