Back to board
coditude·4 months ago

Job Title: Senior AI Agent & Orchestration Engineer

HybridFull-timeMid · 5-8 yearsH1B likely

About this role



Job Title: Senior AI Agent & Orchestration Engineer

Location: [Remote / Office Location]
Role Type: Full-Time / Lead Role Overview
We are looking for an Autonomous AI Engineer to lead the development, orchestration, and production-level observation of our next-generation AI agent systems. You won’t just be calling APIs; you will be architecting stateful, multi-agent workflows that solve complex, real-world problems with minimal supervision.
The ideal candidate is a "Builder-First" engineer who thrives in ambiguity, treats "Vibe Coding" as a starting point but not the finish line, and takes full ownership of the lifecycle—from initial spec to production observability.

### Technical Skill Stack

1. Orchestration & Agentic Frameworks (The "Brain")

  • Libraries: Mastery of LangGraph (for stateful graphs), CrewAI or Microsoft AutoGen (for multi-agent roles), and DSPy (for programmatic prompt optimization).
  • Concepts: Deep understanding of ReAct loops, hierarchical vs. joint collaboration patterns, and Long-Horizon Reasoning.
  • Protocols: Experience implementing the Model Context Protocol (MCP) to connect agents to local/secure data and tools.

2. Advanced RAG & Data Strategy (The "Knowledge")

  • Advanced RAG: Beyond basic vector search—implementing Parent-Document Retrieval, Self-RAG (CRAG), and HyDE.
  • Vector Infrastructure: Proficiency with pgvector (PostgreSQL), Pinecone, or Qdrant.
  • Hybrid Search: Skill in blending semantic (vector) search with keyword (BM25) search for high-precision retrieval.

3. Observability & AgentOps (The "Eyes")

  • Tools: Hands-on experience with LangSmith, Arize Phoenix, or Langfuse for distributed tracing.
  • Evaluations (Evals): Ability to build automated "LLM-as-a-Judge" benchmarks and regression tests to measure agent accuracy and hallucination rates.
  • Production Safety: Implementing cost/latency guardrails and human-in-the-loop (HITL) triggers.

4. Core Engineering (The "Foundation")

  • Languages: Expert-level Python and/or TypeScript.
  • Mobile/Backend Integration: Familiarity with integrating AI services into iOS/Android environments and high-performance backend APIs.
  • Environment: Comfortable in a Linux (Ubuntu) dev environment using Docker, Kubernetes, and local LLM execution (Ollama/LM Studio) for private testing.

### Key Responsibilities

  • Solution Architecting: Take a vague business problem (e.g., "Automate healthcare data anonymization") and design the full agentic workflow.
  • Autonomous Problem Solving: Identify "Prompt Drift" or "Integration Tax" bottlenecks and proactively build tools or protocols to fix them.
  • Orchestration: Build and maintain complex agent graphs that involve tool-calling, API integration, and multi-step reasoning.
  • Observation: Own the production reliability. If an agent fails or enters an infinite loop, you are the one who has already built the trace to see why.

### Soft Skills & Mindset

  • Radical Autonomy: You are expected to find solutions where documentation is thin.
  • Spec-Driven Mindset: You value structured specifications over "trial and error" prompting.
  • Domain Curiosity: Eagerness to dive deep into vertical domains (like Healthcare or Finance) to understand the data context.

Similar jobs