Responsibilities
- Orchestration of Agentive Flows: Build loop logics (Think, Act, Observe) for complex Supply Chain tasks (stockout analysis, recalculating Sell‑In) using frameworks such as LangGraph, CrewAI, Agno or Bedrock AgentCore.
- Implementation of MCP Servers: Develop and expose tools via Model Context Protocol, creating secure bridges (APIs) between LLMs and PostgreSQL, Iceberg databases and Rule Engines.
- Reasoning and State Engineering: Manage short‑ and long‑term memory and implement Thought Signatures to ensure cohesion in multi‑step executions (Gemini 3 and Claude 3.5).
- Tool/Function Calling Optimization: Design strict contracts using JSON Schemas, implementing fallback and retry logic to mitigate hallucinations in external calls.
- AgentOps Discipline: Ensure full AI observability (tracing, latency, token consumption, and decision paths) via LangSmith, Phoenix or AWS CloudWatch.
- Guardrails and Security: Configure automated barriers (e.g., Amazon Bedrock Guardrails) to intercept hallucinations and block unauthorized actions on master data.
Requirements
- Software Engineering: Strong foundation in Python or TypeScript, with focus on microservices, asynchronous code, and clean architectures.
- Mastery of Agents and Tool Use: Deep experience connecting LLMs to tools via native Function Calling (OpenAI, Anthropic, Gemini API).
- Frameworks and MCP: Knowledge of orchestrators (LangGraph, LlamaIndex Workflows, Semantic Kernel) and the Model Context Protocol standard.
- Context Management: Technical experience in memory persistence and handling massive context windows without performance degradation.
- Non‑Deterministic Systems: Ability to create Evals (assessments) and automated tests for probabilistic software, ensuring pre‑deploy reliability.
Core Competencies
Demonstrates expertise in implementing autonomous reasoning flows and managing non-deterministic systems, with a strong foundation in Python or TypeScript for microservices and asynchronous code. Proficient in orchestrating agentive flows and ensuring AI observability through advanced tools and frameworks.
