Claude Code and AI-Assisted Data Engineering

• Why professional data engineers use AI coding assistants daily. This isn't "cheating," it's the current baseline expectation on real teams
• What changes and what doesn't: you're still responsible for correctness, cost, and security. The assistant writes, you validate
• The core discipline: read every diff, understand every query, never ship what you can't explain

You'll learn to: use AI assistance the way a senior engineer does: as leverage, not as a replacement for understanding.

• Setting up and navigating a real project with Claude Code
• Core workflow: plan → edit → run → review diff → commit
• Effective prompting for engineering tasks: giving the right context, scoping requests, iterating

You'll learn to: operate Claude Code as your daily driver for writing and reviewing code, not just as a chat window.

• For every service in Track 2 (GCS, BigQuery, Cloud Run, Workflows, Terraform, etc.), you'll first learn what it is and why it's used, then use Claude Code to write and adapt the actual Python, SQL, and Terraform for it
• Applying this to real pipeline work: writing ingestion scripts, transformation SQL, validation logic, and infrastructure code with Claude Code assisting throughout
• Debugging real pipeline failures with AI assistance: reading logs, root-causing errors, verifying the fix instead of blindly accepting it

You'll learn to: build the exact same capstone pipeline from Track 1, faster and more confidently, using AI as your pairing partner at every layer.

• What an agent is, and when to delegate a task to one instead of doing it inline
• Specialized agents for research, code review, and multi-step implementation work
• Running work in parallel vs sequentially, and when each makes sense

You'll learn to: break down a large engineering task and delegate pieces of it the way a tech lead delegates to a team.

• What MCP is and why it matters: connecting Claude Code to external tools and data (Jira, Confluence, GitHub, databases, cloud APIs)
• Using MCP-connected tools inside real engineering workflows: pulling tickets, reading docs, querying systems without leaving your editor

You'll learn to: extend Claude Code beyond your local codebase into the real tools your team already uses.

• What a "Skill" is: a packaged, reusable set of instructions for a recurring task or workflow
• Using existing skills effectively, and recognizing when a repeated task should become one

You'll learn to: turn your own repeated workflows (a deployment checklist, a code review pass, a report format) into something Claude Code can run consistently.

• How Claude Code remembers context about you, your project, and your preferences across sessions
• What's worth persisting (decisions, feedback, project context) vs what should stay one-off

You'll learn to: work with an assistant that gets more useful over time instead of starting from zero every session.

• How long sessions are handled, and why context degrades if it isn't managed
• Practical habits: scoping conversations, summarizing before switching tasks, keeping signal high in long engineering sessions

You'll learn to: run long, complex engineering sessions (like building your capstone pipeline) without losing coherence halfway through.

• Orchestrating multiple agents toward one larger goal. This directly mirrors the pipeline orchestration concepts from Track 1 (dependencies, sequential vs parallel execution)
• Real scenarios: one agent researching an approach while another implements, or agents reviewing each other's work

You'll learn to: think about AI-assisted work the same way you think about pipeline orchestration: as a system of coordinated tasks, not a single black box.

• Using Claude Code to draft and refine resume bullets from your actual project work (Track 4)
• Drafting LinkedIn "About" sections, Featured post writeups, and outreach messages with AI assistance, then editing for your authentic voice
• Practicing mock interview answers and getting structured feedback

You'll learn to: apply the same AI-assisted workflow to your job search that you use for engineering: faster iteration, still your own judgment in the driver's seat.