Workshops: Agentic Coding for Roles & Industries

Coding Agents for Researchers: Python, R & Reproducible Code

Many researchers write code every day without ever having been trained in software development. Coding agents such as Claude Code, OpenAI Codex or GitHub Copilot can speed up analyses, simulations and data preparation considerably. This workshop shows how to use them while keeping your results reproducible, verifiable and ready for publication.

Who is this workshop for?

  • PhD students and postdocs with analysis or simulation code
  • Scientists in the natural, life, engineering and social sciences
  • Research software engineers and scientific programming teams
  • Research groups working with Jupyter, RStudio or compute clusters
  • Corporate R&D teams running their own data analysis

What you will take away

  • Have a coding agent explain, tidy up and document existing analysis code.
  • Turn a sprawling notebook into a reproducible project with an environment file, version control and fixed random seeds.
  • Check agent-written code with tests against known results instead of trusting it blindly.
  • Generate job scripts for compute clusters without giving the agent your credentials.
  • Handle sensitive research data correctly and disclose AI use in publications.

Agenda

  1. Module 1

    First steps, no prior experience needed

    • What a coding agent is and how it differs from a chatbot
    • Claude Code, OpenAI Codex and GitHub Copilot in VS Code, JupyterLab and RStudio
    • Exercise: have your own analysis code explained and commented
  2. Module 2

    From notebook to reproducible project

    • Moving notebook code into functions and modules
    • Pinning environments, for example with conda, venv or renv
    • Git basics for researchers: versions, branches, undoing changes
  3. Module 3

    Trust, but test

    • Tests against analytical solutions and small synthetic datasets
    • When the agent quietly changes the method: cross-checking statistics
    • Generating figures and tables automatically and traceably
  4. Module 4

    Compute clusters and pipelines

    • Having Slurm job scripts and parameter sweeps generated
    • Workflow tools such as Snakemake or Nextflow at a glance
    • The agent writes, you submit: keeping access separate
  5. Module 5

    Data and good scientific practice

    • Personal and confidential data: what must stay out of prompts
    • Local models for especially sensitive data
    • Documenting AI use and making software citable

Why research code needs its own approach

In product development a bug usually surfaces sooner or later. In research, a wrong parameter or a silently swapped statistical method can go unnoticed for months and end up in a paper. Coding agents write convincing code even when it is wrong in substance.

So the focus is not maximum speed but a way of working in which every change the agent makes can be traced and tested.

Reproducibility as the common thread

By the end of the day your own project has a structure that lets colleagues, reviewers or your future self regenerate the results.

  • Pinned package versions and environments
  • Data kept separate from code, raw data untouched
  • Tests for the core calculations
  • A README with step-by-step reproduction instructions

Sensitive research data

Anything an agent reads may be sent to the model provider. For patient data, interview transcripts or confidential industry data we therefore work with synthetic sample data, business accounts with suitable data settings or locally hosted models. Clearing your project with the data protection officer and ethics committee remains your task; we show which questions to ask them.

Frequently asked questions

Do I need strong programming skills to take part?

No. Basic Python or R is enough. The workshop is designed precisely for researchers who have to write code but never formally learned software development.

Do I have to disclose AI use in my publication?

Requirements differ between publishers, journals and funders. We show how to document AI use so you can meet each set of requirements with little effort.

Can we practise with our own data and projects?

Yes, as long as the data situation allows it. Otherwise we prepare synthetic datasets that mirror your structure, so the exercises remain realistic.

Workshop

Get in touch

Want to speed up your research with coding agents without compromising reproducibility?

Tell us roughly how many people will join and pick dates that work – we will get back to you within 24 hours with a proposal for content, format and schedule.

Prefer to talk first?

15 minutes, free and without obligation – with our AI Trainers and AI Code Engineers.

Book a call for free (opens in new tab) +49 761 519 751 50 (24/7)

Request a training / workshop

Workshop: Coding Agents for Researchers: Python, R & Reproducible Code · 1 day

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy