Generative Artificial Intelligence

Product, operations, and service leaders looking to deploy gen-AI

Data, analytics, and automation practitioners

Solution architects, software engineers, and citizen developers

Compliance, risk, security, and governance partners

Learning, communications, and support teams adopting AI in workflows

Understand how generative models work and where they add value in real processes

Design prompts, chains, and agent workflows that are reproducible and auditable

Apply retrieval-augmented generation with clean data pipelines and guardrails

Use adaptation methods (fine-tuning, instruction tuning, adapters) for domain tasks

Evaluate quality, safety, and cost with clear metrics and test sets

Establish governance for privacy, security, IP, bias, and change management

Foundations of Generative AI

• Model families and use-cases: LLMs, diffusion, transformers; strengths and limits
• From input to output: tokenization, context windows, temperature, decoding strategies
• Capability map: text, code, images, structured outputs, and tool use
• Risks landscape: hallucinations, leakage, bias, prompt injection
• Value framing: pick the right use-case with ROI, risk, and feasibility signals

Prompt Engineering & Workflow Design

• Prompt patterns: role/task context, constraints, examples, evaluation rubrics
• Structured outputs: JSON schemas, validation, and post-processing
• Multi-step chains: summarization → planning → generation → QA
• Agents and tools: calling search, databases, or internal APIs safely
• Ops hygiene: prompt versioning, test cases, and replayable runs

Retrieval-Augmented Generation (RAG)

• Data readiness: cleansing, chunking, embeddings, metadata, and freshness
• Index choices and query strategies; hybrid retrieval for accuracy
• Grounding techniques: citations, reference windows, and confidence cues
• Guardrails: source whitelisting, PII filtering, and policy enforcement
• Hands-on pattern: build a question-answering assistant over private docs

Adaptation & Productization

• When to fine-tune vs. instruct vs. use adapters/LoRA; cost/benefit lens
• Domain adaptation datasets: quality, balance, and labeling guidance
• Integration patterns: apps, BPM/RPA, BI dashboards, and chat surfaces
• Performance and cost: caching, batching, distillation, and model selection
• Change management: enablement, playbooks, and support workflows

Evaluation, Safety & Governance

• Quality metrics: accuracy, faithfulness, completeness, toxicity, bias
• Human-in-the-loop reviews and red-team exercises
• Security: prompt injection defenses, secrets hygiene, output controls
• IP and compliance: copyright, licensing, audit trails, retention policies
• Operating model: roles, approvals, incident handling, and KPI dashboards

Use-case selection and value framing for gen-AI

Prompt engineering, chaining, and agent design

RAG pipelines, data readiness, and guardrails

Adaptation approaches and product integration

Evaluation design, safety practices, and governance setup