Scribd
Upload
34 views2 pages

Take Home Project

The document outlines a take-home project focused on learning three AI frameworks: Google's ADK, Langgraph, and n8n.io. Participants are tasked with identifying a simple problem and implementing an AI agent using these frameworks, emphasizing robust solutions and multi-agent communication protocols. Grading criteria include the value of the idea, functionality, and user interface, with an emphasis on understanding the code and utilizing available resources for development.

Uploaded by

shifeau
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
34 views2 pages

Take Home Project

The document outlines a take-home project focused on learning three AI frameworks: Google's ADK, Langgraph, and n8n.io. Participants are tasked with identifying a simple problem and implementing an AI agent using these frameworks, emphasizing robust solutions and multi-agent communication protocols. Grading criteria include the value of the idea, functionality, and user interface, with an emphasis on understanding the code and utilizing available resources for development.

Uploaded by

shifeau
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Take home project

Learn about these 3 frameworks:

Google's ADK
Langgraph
[Link]

Come up with a simple idea/problem and implement an AI agent to address it


using these 3 frameworks.

Teach me something I don't know about AI. Consider me an investor and


you're pitching your idea/solution to me, try to impress.

Grading will factor these points:

idea or problem you're solving brings value not just money wise
the solution you're building is robust (no errors running it)
your solution makes use of multi subagents architecture and A2A protocol to communicate
between them
you're using MCP protocol to communicate between an agent and its tools
you have two weeks, the more features you add and the nicer the UI is the better, however
functionality and robustness are more important.

Tips and references:

Don't wait to find an idea first, start with the learning process and you will get inspired by
what you can build
you can implement the same idea with each of these framework (the goal here is to learn
how to use each framework well) or you can have different ideas if you wish
Don't use complicated ideas, start very simple and build on top of it. A simple but working
idea is better than a complex crazy one that has many flaws.
n8n is good for doing a quick POC of MVP, but the others are more coding heavy and more
important to be good at so give them extra effort.
use an agentic IDE like: Cursor, Windsurf .. etc (they offer free trials, you can also use a
completely free one like Trae, or student free vscode extensions)
use any references you like online, any AI system, use github repos for inspiration BUT you
have to understand every line of code you're writing/copying
use openrouter to have access to existing powerful yet free LLMs, instead of running locally,
you can also run locally via ollama but it won't be powerful

References:

openrouter models filtered by ones that supports tool calling and ordered by price:
[Link]
high&fmt=cards&supported_parameters=tools
Intro to Langgraph course: [Link]
Marketpalces for existing MCP servers (libraries for AI agents) you can use if you want:
[Link]
[Link]
[Link]
Langgraph official docs: [Link]
Google ADK docs: [Link]
ADK agent samples: [Link]
All the tools have a free version, you can use that. e.g: use Supabase or Pinecone free
support for VectorDB.
Youtube videos for each framework:
Google ADK: [Link]
Langchain: [Link]
[Link] -> you can search for it yourselves, it's everywhere

You might also like