| Folder | Purpose |
|---|---|
/Simple RAG Templates/ |
Minimal, self-contained notebooks that walk you through classic Retrieval-Augmented Generation pipelines. Each file is a “cookbook” experimenting with a different combination of embedding model (OpenAI text-embedding-3, Hugging Face E5-small, etc.) and LLM back-end (GPT-4o, Claude-3, Gemma 1.1, …). Clone, swap keys, run – perfect for first-week prototypes. |
/Agentic RAG Templates/ |
Step-up examples that introduce agent patterns: • Single-agent scripts that plan → retrieve → generate in one loop. • Manager-agent + specialist set-ups where a router LLM delegates to tools (function-calling API) like SQL, Pandas, or vector search. 📝 Highlight: university_rag_pipeline.ipynb shows why numeric CSV columns don’t belong in a vector DB. It spins up a Pandas-helper agent for exact cost-of-attendance queries and a semantic agent for fuzzy “culture” questions—all in ~200 lines of plain Python (no LangChain). |
/Program Resources/ |
PDFs, slides, and week-by-week deliverable checklists mirrored from the master Project Cycles Notion board. |
Project Cycles is AISC’s quarterly build program where small squads spend six weeks shipping an AI agent—culminating in a live ✨ Demo Day ✨.
More context & the full curriculum: https://aisc-uw.notion.site/spr25-resources
Spring 2025 Cohort Showcase: https://aisc-uw.notion.site/pc-spr25
- 90 + students need a safe launchpad ⇢ we give them plug-and-play templates.
- Mixing no-code → full-code tracks means every skill level is welcome.
- Agent techniques (RAG, function-calling, tool routing) so teams start with best practices, not boilerplate spaghetti.