Pathfinder AI: Building a Career Coach
By Alison Morano
In a world where AI is rapidly reshaping the workforce, the tools we use to build our careers should evolve too.
That's what inspired Pathfinder AI — a generative AI-powered career coach designed to help job seekers generate personalized resume summaries, identify matching job patterns, and build a sharper narrative around their professional identity.
This project was developed as part of the Google Generative AI Capstone Challenge using Kaggle and Vertex AI. My goal: create a useful, real-world product that leverages multiple GenAI capabilities, while pushing myself to learn how modern agent frameworks like LangGraph operate in tandem with Google Gemini.
I also created this took for my own use, as I continue to find a new place for my own skills, toolsets, and background.
The Challenge
The capstone challenge required building a working Kaggle Notebook that:
- Uses at least three Generative AI capabilities
- Solves a real-world problem
- Is publicly accessible and demonstrates technical skill
As someone with a background in political messaging and XR, I wanted something that blended communication, AI, and career strategy. The result: Pathfinder AI.
What Pathfinder AI Does
Think of Pathfinder AI as a personal branding co-pilot:
Understands your resume goals
Generates professional, personalized summaries
Matches resume content to job roles using vector search
Builds narrative patterns that can be reused in applications, interviews, or bios
Tools + Architecture
- Component
- Purpose
- Gemini 1.5 Pro
- LLM used for resume summary generation
- LangGraph
- Agent framework to manage tools + flow
- FAISS + LangChain
- Used for storing and retrieving resume chunks
- Google Generative AI SDK
- Direct access for clean Gemini integration
Key GenAI Capabilities Used
- Prompt Engineering: PromptTemplate for dynamic resume summaries
- Vector Embeddings: Used FAISS + Gemini embeddings for job pattern matching
- Agents + Tool Use: LangGraph-powered retrieval agent that responds to user queries about resume content
A Look Inside the Notebook
- Users provide their name, role, skills, and tone
- Gemini generates a customized resume summary
- A LangGraph agent responds to queries like:
- "Find resume patterns for a mid-level marketing manager with automation experience."
- This is handled via a FAISS vectorstore and Google embedding model.
What I Learned
- LangGraph is more flexible and future-proof than older LangChain agents
- Gemini models require precise versioning but perform extremely well
- Vector search adds surprising value to resume analysis — it's more than just keyword matching
What's Next?
- Pathfinder AI is ready to evolve:
- Streamlit front-end for interactive use
- PDF upload and parsing
- Cover letter generation
- Career roadmap suggestions powered by AI
Try It or Fork It
Want to test Pathfinder AI or remix it for your use case? Download the notebook here (link yours).
If you're working on something similar, I’d love to connect. The future of work deserves tools that empower people — and I believe generative AI is at the heart of it.
Alison Berke Morano
Florida-based strategist, XR explorer, and lifelong learner
LinkedIn
American University MIS '25