Inspiration
Helping people who have been frustrated to do this boring process manually ... now, with the AI power, we can automate this process and spread the same concept to other disciplines.
What it does
From a given .pdf files, in this cases, Resumes, the application is able to suggest the best CVs for a new open job position or just do quick searches on these files in a chat mode.
How we built it
- Upload the files to a Google Cloud bucket
- Create a RAG and index these files
- Create 2 agents: one is expert in CVs matches and other to do general searches
- A master agent is able to orchestrate to which agent should route the user request
- A nice UI with Streamlit to interact with the application
Challenges we ran into
- Multi agent integration with RAGs and Streamlit
- Create the perfect prompts to achieve good results
Accomplishments that we're proud of
- The agent matcher is able to find and suggest fast the best CVs that the HR expert could start taking a look with the reasons and why it discards the others (without reading the all huge of resumes)
What we learned
- Learn about the ADK power to create multi agents
- Create RAGs with Vertex AI Search and index files
What's next for A perfect resume for a given job profile
- Integrate with a MCP like LinkedIn to do quick job searches externally
- Some tests and improvements and ready to GO to Production
Built With
- adk
- cloud-run
- gcp
- google-cloud
- python
- rag
- streamlit
- vertexai
Log in or sign up for Devpost to join the conversation.