Skip to content

Add Plot Ark AI Curriculum Engine — multi-agent pipeline with Tavily + LightRAG + GPT-4o-mini#633

Closed
Schlaflied wants to merge 2 commits intoShubhamsaboo:mainfrom
Schlaflied:add-plot-ark-curriculum-engine
Closed

Add Plot Ark AI Curriculum Engine — multi-agent pipeline with Tavily + LightRAG + GPT-4o-mini#633
Schlaflied wants to merge 2 commits intoShubhamsaboo:mainfrom
Schlaflied:add-plot-ark-curriculum-engine

Conversation

@Schlaflied
Copy link

What this adds

A self-contained tutorial for an agentic curriculum generation pipeline built on top of Plot Ark, an open-source EdTech project.

Folder: advanced_ai_agents/multi_agent_apps/ai_curriculum_engine/

How it works

Three agents run in sequence:

  1. Research Agent (Tavily) — searches academic papers, video lectures, and expert articles for the course topic before generating anything, so the curriculum is grounded in real sources
  2. Knowledge Graph Agent (LightRAG) — ingests retrieved sources and builds a per-course entity-relationship graph; hybrid retrieval extracts key concept relationships that inform module design
  3. Curriculum Agent (GPT-4o-mini) — generates structured modules with Bloom's Taxonomy-aligned learning objectives (level-aware: Beginner → L1-2, Intermediate → L3-4, Advanced → L5-6), resources, and assessments

Files

File Description
ai_curriculum_engine.py Main Streamlit app — all three agents + UI
requirements.txt streamlit, openai, tavily-python, lightrag-hku
README.md Setup, architecture diagram, example output

Run it

pip install -r requirements.txt
streamlit run ai_curriculum_engine.py

API keys: OpenAI + Tavily (enter in sidebar or set as env vars).

What makes this different

Most LLM apps generate course content from a prompt alone. This pipeline researches first, then generates — Tavily finds real academic sources, LightRAG extracts concept relationships, and the LLM uses both as grounding context. The result is curricula with real citations instead of hallucinated references.

@awesomekoder
Copy link
Contributor

Thanks for adding actual code this time. The 3-agent pipeline (Tavily research > LightRAG knowledge graph > GPT-4o-mini curriculum) is an interesting pattern.

However, this PR can't be merged as-is for two reasons:

  1. The PR replaces the entire root README.md (264 lines removed, 265 re-added). This looks like a rebase issue. Your PR should only add your tutorial folder + a single line in the README, not rewrite the whole file.

  2. The tutorial README promotes Plot Ark as the parent project. For this repo, tutorials need to stand on their own without framing them as extracts of another product.

If you resubmit with a clean diff (just your folder + 1 README line) and remove the Plot Ark promotion, this could work.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants