Prerequisite: None
William Henry
Senior Catastrophe Pricing Scientist
Mercury Insurance
AI-assisted code generation can meaningfully speed up development work and
lift team productivity, but only when it's paired with the right habits and
guardrails. Most of the value from Claude Code comes from a small set of
habits: clear project instructions, plan-before-edit for non-trivial work,
and a verification loop Claude can run until all checks pass.
In this hands-on workshop—focused on how to use Claude Code effectively for
code generation, not data analysis—your team will practice those
high-leverage workflows across two consecutive half-days.
Attendees will set up project guidance, run plan-then-implement sessions, fix
bugs and ship changes against tests, and add simple guardrails (permissions,
automation, human review) so agentic coding stays safe enough to use day to
day.
Your team will leave with a practical playbook you can apply on your own
repos.
Your Team Will Learn
-
How to get strong results from Claude Code with a small set of
high-leverage habits and workflows
-
How to set up project instructions and context so Claude works well in
your codebase
-
How to plan before editing on non-trivial work, then implement with Claude
Code
-
How to use tests and other checks so Claude can iterate until the work is
actually done
-
Hands-on practice with common coding workflows in Claude Code, from
exploration and bug fixes through commits and pull requests
-
How to add practical guardrails—permissions, automation, and human
review—so agentic coding stays safe to use day to day
-
How to use advanced Claude Code workflows—including running work in
parallel—and when they are worth it versus everyday habits
Geared To
- Data science and analytics managers
- ML engineering leads and player-coaches
- Technical individual contributors newer to agentic AI tooling
- BI/Analytics professionals seeking practical productivity gains
Pre-requisites
Basic familiarity with Python is helpful but not required. No advanced coding
or AI expertise needed.
Laptop Setup:
Students need a laptop with Claude Code installed and working for the
hands-on labs.
Note on Corporate Laptops:
This course requires installation of Claude Code (Anthropic) as well as the
ability to download lab code and data files. If your corporate laptop
prevents these activities, we recommend you use a personal device instead.
System and Software Requirements:
Windows PC or Mac with:
- A modern browser
- Claude Code installed and signed in (claude.ai/code)
-
A paid Claude plan (Pro, Max, or Team) is strongly recommended for full
agent usage during labs
- Git and a modern terminal
- Python 3.10+ with the ability to create a virtual environment
- Ability to clone GitHub repositories used in the labs
Setup:
Instructions will be available on the seminar site prior to the event. You
must prepare your laptop before the seminar begins.
There is no time allotted in class for laptop preparation.