Head of Artificial Intelligence; Agentic AI Platform

Position: Head of Artificial Intelligence (Agentic AI Platform)
Job Description - Director / Head of Agentic AI Platform & Digital Workers

Job Location - Pune, Maharashtra

Experience - 8 to 12 years

Our client is hiring a Senior Engineering Leader to build from scratch, own, and scale an enterprise-grade Agentic AI Platform (“Agentic AI Fabric”) and a portfolio of production Digital Workers (agents) that execute real workflows safely across enterprise systems.

What this leader will own
- 1) Build and own the platform (Agentic AI Fabric)

- Agent runtime/orchestration patterns for multi-step workflows (state, retries, queues/back pressure, approvals/HITL)
- Tool/action gateway and registry (governed connectors, allow lists, parameter validation, rate limits, rollback)
- Knowledge/RAG layer (ingestion pipelines, provenance, access controls, grounding and traceability)
- LLMOps/evaluations (versioned prompts/tools/policies, regression gates, safe rollouts, rollback)
- Observability and cost governance (telemetry, budgets/alerts, cost-per-transaction, SLOs/runbooks)
- Multi-cloud-ready architecture patterns (AWS first; extend to Azure later)

2) Build and scale agents as products

- Deliver a production-grade MVP agent and expand a portfolio of Digital Workers
- Establish a “Next Agent Kit” to accelerate subsequent agents (templates, tool patterns, eval harness, rollout checklists)

3) Enterprise agent security baseline (must-have)

This leader must set and enforce controls for:

- PII/PHI boundary controls and safe logging/data minimization
- Prompt/tool injection defenses
- Knowledge base poisoning protections (provenance, approvals/versioning, trust filters)
- Tool supply-chain integrity (registry governance, version pinning, scanning/SBOM)
- Privilege escalation controls and step-up approvals / separation of duties

Why this is a senior role

This is not an innovation lab position. It requires:

- shipping production systems under enterprise constraints,
- handling security/compliance scrutiny,
- orchestrating cross-functional teams (engineering, security, SMEs, delivery),
- and building something reusable and scalable (platform + agents).

Ideal candidate profile (what “good” looks like)

Must have

- 8 to 12 years in engineering/product/platform development
- 3+ years leading teams and complex delivery programs
- Track record building enterprise products/platforms (not just projects): multi-tenant SaaS, internal developer platforms, automation/workflow platforms, data/AI platforms
- Strong cloud architecture experience (AWS and/or Azure), including IAM/security, networking, and observability
- Hands-on familiarity with LLM applications (RAG + tool calling + evals/monitoring), and the realities of production deployment

- Experience with

Strongly preferred

Agent frameworks & orchestration

- Hands-on experience with agent/workflow orchestration frameworks such as Lang Graph, CrewAI, Auto Gen, and/or similar (e.g., Llama Index agents), including multi-step tool calling and human-in-the-loop patterns.

Automation/workflow platforms

- Experience with workflow automation / integration platforms such as n8n (or comparable low-code workflow engines) to rapidly compose actions and integrate enterprise systems.

Tool integration standards (optional)

- Familiarity with MCP-style tool servers/tool gateways and agent interoperability concepts (A2A-ready patterns), even if not fully implemented end-to-end.
- Has built or scaled a platform where customers/users rely on it daily (SLOs, on-call, release governance)
- Has shipped AI-assisted or autonomous workflow systems that integrate with enterprise apps (ITSM/CRM/ERP)
- Has worked in regulated contexts (BFSI/healthcare) or delivered audit-grade systems

Nice-to-have

- Familiarity with agent frameworks, interoperability concepts (A2A/MCP patterns), secure tool execution
- Experience packaging offerings / enabling GTM narratives for enterprise stakeholders
- Inference engineering and cost/performance optimization
- Experience building and operationalizing evaluation frameworks for LLM/agent workflows