Position Overview :

We are seeking a visionary and hands-on Agentic AI Architect to lead the technical design, governance, and deployment of our next-generation autonomous AI systems.

In this role, you won't just build chatbots or single-turn prompt wrappers; you will architect enterprise-grade, multi-agent ecosystems that actively reason, decompose complex goals into discrete sub-tasks, utilize APIs, and execute end-to-end business operations.

You will bridge the gap between deterministic software architecture and probabilistic AI behaviors, ensuring our autonomous agents are reliable, secure, stateful, and capable of operating under strict enterprise guardrails.

Key Responsibilities :

1. Core Agentic System Architecture :

- Design and implement scalable multi-agent and hierarchical orchestration patterns (e.g., Critic-Actor, Supervisor-Worker, and decentralized collaboration models).

- Architect stateful, deterministic, and fault-tolerant agent execution graphs, ensuring systems can pause for human-in-the-loop (HITL) overrides and gracefully recover from model hallucinations or API failures.

- Establish advanced context engineering and grounding strategies using short-term, long-term, episodic, and Graph RAG (Retrieval-Augmented Generation) memory architectures.

2. Enterprise Tooling & Integration :

- Design and build secure server environments utilizing Model Context Protocol (MCP) to expose database schemas, internal enterprise APIs (ERP, CRM, TMS), and cloud infrastructure to autonomous agents.

- Define semantic contracts and standardized schemas that allow agents to autonomously call backend tools, manage structured outputs, and reason over multi-modal inputs.

- Establish API boundaries and routing logic to optimize latency, tokens, and compute costs across diverse frontier and open-source models (e.g., Claude, GPT, Llama).

3. AgentOps, Observability & Guardrails :

- Implement production-grade AgentOps framework to monitor multi-turn conversations, loop detection, tool-invocation failures, and downstream drift.

- Design and enforce cognitive guardrails, semantic firewalls, and alignment safety checks to prevent prompt injections, goal-hijacking, and unauthorized tool execution.

- Build rigorous data evaluation pipelines to measure agent accuracy, reasoning paths, alignment, and cost-to-outcome metrics.

Required Technical Skills & Experience :

- Agentic Orchestration: Proficient with cutting-edge frameworks like LangGraph, CrewAI, Autogen, Google ADK, or OpenAI Agents SDK.

- Protocols & Tooling: Deep understanding of Model Context Protocol (MCP), Vector Databases (Pinecone, Milvus, Weaviate), and Knowledge Graphs (Neo4j).

- AI/ML Core: Mastery of advanced prompting techniques, context window management, function-calling mechanics, and standard data science frameworks (Python, PyTorch).

- Cloud Architecture: Extensive experience with cloud-native architectures (AWS EKS, GCP Vertex AI, Azure AKS), serverless systems, and event-driven data streaming (Kafka).

- DevSecOps: Familiarity with automated CI/CD for LLM pipelines, infrastructure as code (Terraform), and LLM testing/evaluation tools (Ragas, TruLens).