Workshops: Agentic Coding
Agentic Coding for Data Science & Analytics: Workshop
Data teams spend a lot of time on data preparation, SQL and notebook boilerplate. This workshop shows data scientists and analytics engineers how AI agents speed up that work and how to keep numbers correct and analyses reproducible along the way.
Who is this workshop for?
- Data scientists
- Data analysts and BI developers
- Analytics engineers (dbt, SQL)
- Data engineers
- ML engineers
What you will take away
- Have agents write and validate SQL queries and dbt models against your schema.
- Use agents for data cleaning, exploratory analysis and visualisation in Python.
- Turn notebook experiments into clean, tested code with agent help.
- Build in plausibility checks so wrong numbers do not slip through.
- Know the data protection rules for using agents with company data.
Agenda
Module 1
Agents for data work
- Claude Code, OpenAI Codex, GitHub Copilot and Cursor in data workflows
- Notebooks vs. scripts vs. pipelines
- Context: schema, data catalogue and metric definitions
Module 2
SQL and modelling
- Exercise: have queries written and checked against your data model
- dbt models, tests and docs with agents
- Read-only database access via MCP
Module 3
Analysis and visualisation
- Exploratory analysis with pandas or Polars
- Creating and iterating on charts
- Cross-checking results: samples, totals, reconciliations
Module 4
From notebook to pipeline
- Refactoring into modules and functions
- Data tests and validation
- Orchestration and repeatability
Module 5
Using data safely
- Which data may enter the agent's context
- Pseudonymisation and sampling
- Enterprise plans and GDPR
The particular danger: plausible but wrong numbers
With code, bugs often show up as crashes. With analysis they do not: a wrong join or missing filter returns a number that looks plausible and is still wrong. Agents make these mistakes just like people, only faster.
That is why a focus of the workshop is verification: metric definitions as context, automated data tests and reconciliations the agent runs itself.
What you take away
You work with your own data stack or a comparable practice dataset.
- Project rules for your data repository
- Analysis prompt templates with built-in checks
- A review checklist for data analyses
- Guidelines for handling sensitive data
Format
Remote or on-site, in English or German, delivered by neooutput AI Code Engineers. Optional follow-up coaching.
Frequently asked questions
Isn't this the same as data analysis in ChatGPT?
Chat-based analysis is useful for one-off questions. Agents work inside your repository with your schema, tests and version control, making the work reproducible and team-ready.
Can we use real customer data?
We recommend synthetic, anonymised or pseudonymised data and enterprise plans with suitable contract terms. We clarify what is permissible in your case.
Which languages do you cover?
The focus is SQL and Python. R or Scala can be included on request.
Workshop
Get in touch
Want agents to speed up your analyses without getting the numbers wrong?
Tell us a little about your team – we will get back to you within 24 hours with a proposal for content, format and dates.
Prefer to talk?
More workshops
- Workshops: Agentic CodingAgentic Coding for DevOps & Infrastructure as Code: Workshop
- Workshops: Agentic CodingBuilding MCP Servers: Hands-On Workshop for Dev Teams
- Workshops: Agentic CodingSpec-Driven Development with AI Agents: Workshop
- Deploy & App StoresFrom Cursor to Google Play: Publishing Your Android App
