Job Description

Job requisition ID :: 95305
Date: Jan 12, 2026
Location: Pune
Designation: Associate Director
Entity: Deloitte Touche Tohmatsu India LLP

Responsibilities

Design Next-Generation Agentic AI Architecture

Develop advanced architectural strategies for Agentic AI systems and multi-agent frameworks. Define and implement architectures that enable autonomous decision-making, planning, reasoning, and tool-use capabilities using LLM-driven agents. Architect complex agent hierarchies, coordination mechanisms, memory systems, safety layers, and interaction protocols.

Orchestrate Autonomous Agents & AI Workflows

Design, deploy, and optimize agentic pipelines including task-planning agents, data-retrieval agents, evaluation agents, and self-improving multi-agent systems. Integrate reasoning engines, vector memory, RAG pipelines, long-term memory stores, and dynamic orchestration layers to establish scalable autonomous workflows.

Architect Enterprise-Grade Agent Platforms

Own the architectural vision for agentic platforms such as PepGenX Agent Framework. Drive build-versus-buy evaluations, vendor assessment, and SaaS integration strategies. Ensure architectural cohesion, low coupling, high modularity, and reusability across agent components, sub-systems, and orchestration layers.

Lead Architectural Alignment & Governance

Conduct architecture alignment workshops, contribute to enterprise pattern libraries, and ensure strict adherence to agentic AI design principles. Tag, classify, and onboard new agent components for reuse--accelerating delivery and reducing operational costs. Provide architectural guidance across cross-functional teams including Data Engineering, DS/ML, and Platform Engineering.

Build Enterprise Agent Tooling & Reuse Strategy

Define standards for reusable agent components, templates, orchestrators, evaluators, and interaction modules. Establish governance systems that ensure consistent design patterns, interoperability, and safe scaling across business units.

Implement Agentic LLM Operations (LLMOps)

Lead LLMOps for autonomous agent systems, including observability, drift detection, safety guardrails, memory optimization, and dynamic agent adaptation. Optimize performance across inference, reasoning depth, agent communication, and long-context processing.

Drive Advanced Agent Capabilities

Architect and implement:

Multi-agent collaboration frameworks Recursive self-correction and chain-of-thought verification Autonomous task decomposition and planning Tool-use, API execution, and workflow automation Reflection, evaluation, and self-improvement loops

Ensure enterprise agents operate with reliability, safety, scalability, and compliance with AI governance frameworks.