As an Agent Orchestration Engineer at S&M, you will be the Conductor of our autonomous AI systems. You will design and implement multi-agent architectures that allow specialized AI agents to collaborate, reason, and execute complex workflows across enterprise functions like BFSI, Healthcare, and Retail. Your goal is to move beyond simple chat interfaces to build Action-Oriented Agents that integrate deeply with client APIs and legacy databases.
Key Responsibilities- Agentic Architecture Design: Develop multi-agent workflows using patterns like Sequential, Parallel, and Manager-Worker to solve complex enterprise problems.
- Orchestration Framework Implementation: Build and maintain agent logic using frameworks such as LangGraph, CrewAI, AutoGen, or Semantic Kernel.
- System Integration: Connect AI agents to enterprise systems (SAP, Salesforce, EHRs) via robust tool-calling and API integrations.
- Memory & Context Management: Design persistent memory layers so agents can remember user preferences and past interactions across long sessions.
- Behavioral Guardrails: Implement safety protocols (e.g., NeMo Guardrails) to ensure agents operate within legal, ethical, and operational boundaries.
- State Machine Management: Handle Human-in-the-loop (HITL) checkpoints for high-risk autonomous actions (e.g., approving a $100k loan).
- Performance Monitoring: Use observability tools like LangSmith or Arize Phoenix to debug agent hallucinations and optimize token consumption.
Required Technical Skills- Languages: Advanced proficiency in Python (asyncio, Pydantic) and experience with TypeScript/Node.js.
- Frameworks: Hands-on experience with LangChain/LangGraph, CrewAI, or Microsoft AutoGen.
- Models: Deep understanding of LLM reasoning capabilities (GPT-4o, Claude 3.5 Sonnet, Llama 3) and how to leverage Function Calling and Structured Outputs.
- Data & Memory: Experience with Vector Databases (Pinecone, Weaviate, Milvus) and Graph Databases for knowledge retrieval.
- Deployment: Familiarity with containerization (Docker/K8s) and cloud-native AI services (AWS Bedrock Agents, Azure AI Studio).
