How vinayparmar555/learn_agenticai works
This is a hands-on Agentic AI learning repository: separate examples cover prompting, tool use, retrieval-augmented generation, memory, graph workflows, and structured outputs. The clearest end-to-end workflow is RAG: index documents, retrieve and optionally rerank context, then generate an answer; a queued variant accepts a query, schedules processing, and returns a job result. The README also describes a FastAPI todo app, but the tree shows only its frontend files, so its backend flow is not mapped. Edges are limited to excerpted code or README-documented capabilities; unsampled implementations are represented without guessed internals.
Main entry points: retreival.py RAG Retrieval · raw_openai_func_call.py Tool-calling Agent · main.py RAG API · rq_client.py RQ Queue Client
Components 25
Agent Patterns
- LLM Setup
llm.py - Prompting Patterns
CoT.py - Tool-calling Agent
raw_openai_func_call.py - Agent Tools
raw_openai_func_call.py
RAG Systems
- Document Indexing
Rag/indexing.py - RAG Retrieval
Rag/retreival.py - Hybrid Search
Rag/hybrid_search.py - Qdrant Vector Store
- Cohere Reranker
- RAG Evaluation
Rag/retreival.py
Queued RAG
- RAG API
rag_queue/main.py - RQ Queue Client
rag_queue/clients/rq_client.py - Query Worker
rag_queue/queues/worker.py - Worker Runner
rag_queue/clients/run_worker.py - Redis
Agent Workflows
- Graph Chat
langgraph/chat.py - Conditional Chat
langgraph/chat_2.py - Checkpoint Chat
langgraph/chat_checkpoint.py - Memory Agent
memory_agent/mem.py - Memory Store
Schemas and App
- Output Schema
schema/output.py - Todo Frontend
todo_app/script.js
Other components
- Learner / User
- LLM Providers
- Web Search
Connections 30
- Learner / User → Tool-calling Agent: submits prompt
raw_openai_func_call.py - Tool-calling Agent → Output Schema: validates response
raw_openai_func_call.py - Tool-calling Agent → Agent Tools: dispatches tool
raw_openai_func_call.py - Agent Tools → Web Search: searches web
raw_openai_func_call.py - Tool-calling Agent → LLM Providers: requests completion
raw_openai_func_call.py - LLM Setup → LLM Providers: configures clients
README.md - Learner / User → Document Indexing: provides documents
README.md - Document Indexing → Qdrant Vector Store: indexes vectors
README.md - Learner / User → RAG Retrieval: asks question
Rag/retreival.py - RAG Retrieval → Qdrant Vector Store: searches vectors
Rag/retreival.py - RAG Retrieval → Cohere Reranker: reranks results
Rag/retreival.py - RAG Retrieval → LLM Providers: generates answer
Rag/retreival.py - RAG Retrieval → RAG Evaluation: evaluates answers
Rag/retreival.py - Hybrid Search → Qdrant Vector Store: stores hybrid vectors
Rag/hybrid_search.py - Learner / User → RAG API: submits query
rag_queue/main.py - RAG API → RQ Queue Client: enqueues job
rag_queue/main.py - RAG API → Query Worker: schedules processing
rag_queue/main.py - RAG API → RQ Queue Client: fetches job result
rag_queue/main.py - RQ Queue Client → Redis: connects queue
rag_queue/clients/rq_client.py - Worker Runner → RQ Queue Client: uses connection
rag_queue/clients/run_worker.py - Query Worker → Qdrant Vector Store: searches chunks
rag_queue/queues/worker.py - Query Worker → LLM Providers: generates answer
rag_queue/queues/worker.py - Learner / User → Graph Chat: starts chat
README.md - Learner / User → Conditional Chat: starts workflow
langgraph/chat_2.py - Conditional Chat → LLM Providers: calls model
langgraph/chat_2.py - Learner / User → Checkpoint Chat: starts chat
README.md - Learner / User → Memory Agent: enters query
memory_agent/mem.py - Memory Agent → Memory Store: searches and saves
memory_agent/mem.py - Memory Agent → LLM Providers: generates reply
memory_agent/mem.py - Learner / User → Todo Frontend: uses todo app
README.md