mcp-langchain-agents-with-user-context-api

1. tool

    jobsearch agent 
    news agent  
   

2. agentic ai tool (langchain) - MCP structure

    agent determines based on user query which tool to invode 
   

3. rag to store & retrive user level historical conversation for context

4. fastapi

    user management
    concurrency 
    user level context reterival 
   

5. tested concurrency of the rest-api

6. Note MCP is not implemented here from mcp package, but under folder "additional_codes_mcp_like_implementation", a fast api server is created mcp_job_search and the same is called in code llm_microservice. This is same concept as MCP

note:

ai tools were leveraged and used as assistant to create codes in this repo 
databricks was used as llm were already provided & served as endpoint 
alternative approach: 
    1. deploy fastapi in AKS instead of long running job clusters -- that is more optimal for hosting microservice 
    2. creata a custom endpoint in databricks model serving

next steps:

create a approach for evaluation 
    1. evaluate at each process 
    2. overall functional evaluation