14 teaching weeks · 28 hours · life-science students

AI Agents in Python

Applied LLMs for Life Sciences · From vibe coding to reliable agentic applications

Course contact: william.droz@idiap.ch

Download course materials (.zip) · Extract the folder, then follow the computer setup guide.

Currently available: Weeks 1–5. Later weeks are shown below and will be released during the course.

Weekly slides

  1. Week 01 First PydanticAI chat and LLM literacy Open slides PDF view
  2. Week 02 Vibe coding for exploration Open slides PDF view
  3. Week 03 A tiny app from a short brief Open slides PDF view
  4. Week 04 Ask a dataset, build the plots Open slides PDF view
  5. Week 05 Connect Four: models, decisions, and MCP Open slides PDF view
  6. Week 06 Design verification and correction loops Not yet released
  7. Week 07 Retrieval and agentic RAG Not yet released
  8. Week 08 Code Mode and controlled-tool comparison Not yet released
  9. Week 09 MCP queries, plots, and artifacts Not yet released
  10. Week 10 Evaluate an agent against evidence Not yet released
  11. Week 11 Review and approve an agent's output Not yet released
  12. Week 12 Testing, evaluation, and observability Not yet released
  13. Week 13 Human-in-the-loop application Not yet released
  14. Week 14 Integrative reliability review Not yet released

Course resources

ASSESSMENT

Evaluation project

Three weekly blocks and rubric

Build a reproducible, human-reviewed scientific report assistant step by step in Weeks 12–14.

  • Week 12: reproducible analysis and candidate figures
  • Week 13: confirmed evidence and draft preview
  • Week 14: human approval and reviewed Typst/PDF export
Not yet released
OPTIONAL REFERENCE

Optional workflow patterns

From Week 6

Compare common workflow patterns after designing a bounded verification loop.

  • Workflow and agent design choices
  • Routing, chaining, parallel work and evaluation
  • Reference material without an additional lab
Not yet released