AI Lead - Agentic & Conversational AI

** Introduction*
* At IBM, we believe technology shapes the world. We're a catalyst for that innovation. We're driving change that improves businesses, society, and the human experience. Our Marketing, Communications & Corporate Social Responsibility (MCC) team tells this story. We shape IBM's brand, capture attention in the market, and share our perspective with clients, partners, the media, and fellow IBMers. On our team, you'll work with bright, collaborative minds who bring passion and creativity to everything they do.

You'll be part of a culture built on openness, trust, and teamwork. Where your ideas matter and your growth is supported. Join us, and help bring innovation to life.

** Your role and responsibilities*
* About the Role

We are building a small, elite AI strike team embedded in the Marketing, Communications, and CSR (MCC) Strategy & Operations team. The mission is to move fast, explore what's possible with agentic and conversational AI, and prove value quickly-before anything is scaled or productionized elsewhere.

This role leads that squad.

You will be a hands‑on AI leader with deep expertise in Large Language Models, conversational AI, and agent design. You will guide rapid experiments, build compelling proof‑of‑concepts, and help senior marketing leaders understand how AI can meaningfully change how work gets done.

This is not a traditional AI platform role, a reporting function, or a research lab. It is a high‑judgment, rapid‑execution role optimized for learning, momentum, and decision‑making.

This Role IS

A hands‑on leader for a small, elite AI delivery team

Focused on agentic AI, conversational experiences, and copilots

Oriented around workflow transformation, not tools for their own sake

Optimized for speed, experimentation, and proof‑of‑value

Embedded directly in MCC Strategy & Operations Team

Closely partnered with senior Marketing, Communications and CSR leaders

This Role Is NOT

A dashboarding or reporting function

A long‑term AI platform or MLOps ownership role

A research‑only or theory‑heavy AI role

A large team or program management position

Core Responsibilities

1. Leadership & Execution

Lead a small, high‑trust AI squad focused on rapid delivery

Set clear, time‑boxed missions with explicit learning or go/no‑go outcomes

Balance hands‑on building with technical direction and coaching

Operate with a "build → test → learn → decide" mindset

Optimize for progress and insight, not polish or permanence

2. Conversational AI & LLM Expertise (Core Focus)

Apply deep understanding of LLM capabilities and constraints, including:

Tokenization and context windows

Latency and cost tradeoffs

Failure modes and hallucination risks

Design multi‑turn, high‑quality conversational experiences

Manage conversation state, context, and memory deliberately

Create flows that balance strong UX, deterministic logic, and LLM flexibility

3. Query Understanding, Routing & Agent Selection

Design systems that translate user input into confident routing decisions

Apply intent detection, semantic similarity, and hybrid triage approaches

Decompose complex requests into actionable subtasks

Select and orchestrate agents based on role, context, confidence, and history

Build fallback paths, confidence scoring, and uncertainty handling

Implement guardrails to prevent unsafe, incorrect, or misleading behavior

4. Agent Design & Orchestration Patterns

Lead hands‑on experimentation with proven agent patterns, including:

Router / Dispatcher

Planner-Executor

Re Act

Tool‑calling agents

Retrieval‑Augmented Generation (RAG)

Make pragmatic choices between prompt‑driven and code‑driven controls

Optimize for predictable, explainable behavior, not full autonomy

Design agents that are credible and trusted in an enterprise marketing context

5. Prompt Engineering & Rapid Iteration

Develop advanced system, developer, and user prompt architectures

Create prompt templates for routing, planning, classification, and tool use

Manage prompt versioning and experimentation

Use constraints, grounding, and structured outputs to reduce hallucinations

Iterate quickly based on live usage and feedback

6. Lightweight Governance & Enterprise Awareness

Ensure experiments consider safety,…