Role Summary

- Lead the design and deployment of end-to-end agentic AI systems, owning the full architecture stackfrom knowledge curation to cognitive reasoning to autonomous execution.
- This role is accountable for building a multi-layered AI system architecture where agents:
- Understand enterprise context (knowledge layer)
- Reason and plan (cognitive layer)

- Execute actions (agentic layer)
- Continuously improve (feedback + performance layer)
- You are not building isolated AI features—you are architecting an enterprise AI operating system.
- Architect and operationalize the full agentic AI stack
- Build reusable AI system layers and components
- Enable scalable, governed, high-performance autonomous enterprise workflows
- Expanded Responsibilities: Full Agentic AI Stack Ownership

1. Knowledge Curation & Semantic Layer
- Define strategy for enterprise knowledge ingestion, curation, and structuring

- Build pipelines for

- Structured + unstructured data
- Documents, APIs, real-time streams

- Establish

- Metadata frameworks
- Architecture for consuming ontologies / semantic models
- Ensure knowledge is AI-consumable, contextual, and continuously updated
- Outcome: A trusted, dynamic enterprise knowledge foundation
- Design a cognitive catalog that indexes:

- Agents

- Tools/APIs

- Skills and capabilities
- Enable discoverability and reuse of:

- Prompts

- Workflows

- Models
- Build a system where agents can discover and invoke other agents/tools
- Outcome: A self-service, composable AI capability layer

3. Decisioning Framework
- Creative / Generative Intelligence

- LLM orchestration for

- Content generation

- Hypothesis creation

- Natural language reasoning




- Manage multi-model strategy (cost vs performance vs specialization)

- Logical / Deterministic Intelligence
- Rule engines, mathematical reasoning, workflow logic

- Integrate with AI/ML models

- Compliance

- Model accuracy

- Hybrid AI systems combining
- LLM reasoning + programmatic control
- Outcome: Balanced creativity + reliability in AI decisioning

4. Agentic Layer (Autonomous Systems Design)
- Architect:
- Single-agent and multi-agent performant systems
- Hierarchical and collaborative agent models

- Define
- Planning, memory, and execution loops

- Task decomposition and coordination

- Enable agents to
- Take actions across enterprise systems

- Learn from feedback
- Outcome: Production-grade autonomous workflows

5. Agentic Integration Layer
- Design integration with:
- Enterprise applications (CRM, ERP, HR systems)

- Data platforms and APIs
- Build secure action frameworks for agents:

- API orchestration
- Event-driven architectures
- Ensure agents can execute real business transactions
- Outcome: AI moves from insight action

6. Data Mesh & Distributed Data Architecture
- Align agentic systems with data mesh principles
- Enable domain-driven data ownership

- Ensure

- Data discoverability

- Data product standardization

- Integrate with platforms like

- Databricks

- Snowflake
- Outcome: Scalable,



domain-aligned data foundation for AI

7. Governance, Security & Control Framework
- Define governance for:
- Autonomous decision-making

- Data access and privacy

- Implement
- Role-based access controls for agents
- Human-in-the-loop mechanisms

- Audit trails and explainability
- Ensure compliance with enterprise and regulatory standards
- Outcome: Trusted and controllable AI systems

8. Performance, FinOps, Observability & Optimization
- Define and track:

- Task success rate

- Agent autonomy levels

- Cost per execution

- Latency and throughput

- Build observability stack for

- Agent behavior

- Failure modes

- Optimize using

- Feedback loops

- Continuous learning systems
- Outcome: Reliable, productive, and scalable AI operations

9. Platform Engineering & Reusable Frameworks
- Build Agentic AI development platform with reusable:
- Agent frameworks & templates

- Agent repository and discoverability

- Orchestration layers

- Governance layers

- SDKs and accelerators

- Productize Agentic capabilities into
- Client-facing offerings

- Repeatable solutions
- Outcome: IP-led AI engineering business

Must-Have
- 15+ years in distributed systems, AI/ML, or platform engineering
- Deep hands-on experience building:
- LLM-based systems
- Agentic or workflow automation platforms
- Proven experience delivering enterprise-scale AI systems in production

- Critical Differentiators
- Has architected multi-layer AI systems (not just apps)

- Experience with
- Knowledge systems (RAG, ontologies)
- Multi-agent orchestration

- AI governance frameworks
- Strong engineering depth + business acumen

📌 AI Solution Architect (Chennai)
🏢 CEI AI
📍 Chennai