MODULE 21

Agentic AI for Data Analysis, Data Engineering, and AutoML

Design agentic workflows for analysis, SQL, pipelines, AutoML, and analytics operations with tools, state, guardrails, evaluation, and approval.

Advanced 24 min

From assisted AI to agentic AI for data work

From AI-Assisted to Agentic AI for Data Work on GinnyTech: choosing when AI assistance is enough and when a governed agentic workflow with controls, ownership, and revisable outputs is needed.

Prerequisites: 1

Go to lesson

Advanced 30 min

Tools, state, memory, handoff, and guardrail

Tools, state, memory, handoff, and guardrails on GinnyTech: define which tools an agent can use and which actions require review with controls, ownership, and revisable outputs.

Prerequisites: 1

Go to lesson

Advanced 29 min

Agentic data analysis with human controls

Agentic data analysis with human controls on GinnyTech: establishing where the agent can proceed and where it must stop for approval with controls, ownership and reviewable outputs.

Prerequisites: 1

Go to lesson

Advanced 31 min

Agentic SQL and semantic layer with approval

Agentic SQL and semantic layer with approval on GinnyTech: decide if an agentic query can become a reusable model or remains exploration with controls, ownership, and reviewable outputs.

Prerequisites: 1

Go to lesson

Agentic data engineering: pipelines, runbooks, and incidents - GinnyTech header image with an editorial cosmic visual

Advanced 32 min

Agentic data engineering: pipeline, runbook, and incidents

Agentic data engineering: pipelines, runbooks, and incidents at GinnyTech: establishing which operational actions an agent can suggest or perform during data incidents with controls, ownership, and revisable outputs.

Prerequisites: 1

Go to lesson

Advanced 34 min

Agentic AutoML: planning, training, review, and deploy

Agentic AutoML: planning, training, review, and deploy on GinnyTech: deciding when an agent can start AutoML and when it must ask for human confirmation with controls, ownership, and reviewable outputs.

Prerequisites: 1

Go to lesson

Advanced 33 min

Multi-agent workflow: analyst, engineer, reviewer, owner

Multi-agent workflow: analyst, engineer, reviewer, owner on GinnyTech: design a multi-agent collaboration where each agent has limited and verifiable responsibilities with controls, ownership, and revisable outputs.

Prerequisites: 1

Go to lesson

Observability, evaluations, security, and cost control - GinnyTech header image with an editorial cosmic visual

Advanced 35 min

Observability, evals, security and cost control

Observability, evaluations, security, and cost control on GinnyTech: decide if an agentic workflow is sufficiently observable and secure for recurring use with controls, ownership, and revisable outputs.

Prerequisites: 1

Go to lesson

Advanced 36 min

Case study: agentic system for analytics operations

Case study: agentic system for analytics operations on GinnyTech: design an end-to-end agentic system with clear operational boundaries with controls, ownership, and reviewable outputs.

Prerequisites: 1

Go to lesson

Advanced 20 min

Cheat sheet: agentic AI for data work

Cheat Sheet: Agentic AI for Data Work on GinnyTech: using a checklist to approve, limit, or stop agentic workflows in data with controls, ownership, and revisable outputs.

Prerequisites: 1

Go to lesson

Module connections

Continue the path with these modules

Natural prerequisiteAI for Data Analysis, Data Engineering and AutoML10 lessons

Use generative AI and AutoML to accelerate analysis, pipelines, features, and decision memos without losing control, quality, and ownership.

Same stage of the pathPhilosophical Foundations of Data Analysis20 lessons

Give epistemological, causal, and critical depth to data work, linking evidence, models, risk, and responsibility.

Same stage of the pathDirections in Analytics: Marketing, Product, Finance14 lessons

Connect technical modules to real roles, economic responsibilities, and professional specialization choices.

Previous moduleCourse Overview and Study Method for Data Work8 lessons

Transform the course from catalog to cognitive system, with study method, technical memory, and professional output plan.

Previous moduleData Collection & Tracking Systems8 lessons

Move from isolated tags and tools to reliable, governed, and observable measurement systems.

Previous moduleMetrics, KPI Trees, and Analytical Foundations6 lessons

Learn to read numbers as systems, with KPI trees, unit economics, correct baselines, and causal signal interpretation.