Hidden Paths: Unlocking Unexpected Career Journeys with AI and GraphRAG 🚀

Imagine a world where learning knows no bounds—platforms like Facebook, Udemy, and Coursera offer endless opportunities to gain new skills. Yet, the job market grows more complex every day, and traditional career counseling is often expensive and time-consuming. For those with limited experience, finding the right career path can feel impossible. That’s where Hidden Paths steps in. Inspired by the iconic job progression system in Ragnarok Online—where career paths unfold in a beautifully mapped graph—we set out to create a tool that suggests unexpected yet achievable career paths based on your unique strengths.


🌟 What We’re Building

Our web app harnesses the power of AI and GraphRAG to deliver personalized career recommendations tailored to your skills, interests, and education. Hidden Paths offers two core features:

  • Generate Career Graph 🌳: Input your details—your current profession, skills, interests, and education—and watch as the system crafts a Ragnarok Online-style graph, revealing career paths you might never have considered but are perfectly within your reach.
  • Q&A System 💬: Have questions about careers, skills, or interests? Ask away! Our bot transforms your questions into AQL queries, pulling answers straight from our Graph Database in ArangoDB.

We built this for high school students, university students, professionals seeking a career pivot, job seekers eager to upskill, and even organizations fostering employee growth.


🛠️ How We Made It

To bring Hidden Paths to life, we started by gathering data on professions and their required skills, aiming to create a web of relationships in a graph format. When we hit gaps in the data, we turned to Generative AI to fill them in and clean things up. That data was then transformed into a NetworkX graph and stored in ArangoDB’s Graph Database. Using GraphRAG, we generated dynamic career path graphs, while our Q&A system relied on converting user questions into AQL queries—with a little help from detailed prompts to boost the LLM’s accuracy.


⚠️ Challenges Along the Way

The journey wasn’t without its hurdles:

  • Q&A Hiccups 😓: The AQL queries generated by our LLM often failed to execute, leaving us unable to answer user questions. We tackled this by mapping user inputs to match database terms and feeding the LLM detailed explanations of our data structure.
  • Time Crunch ⏰: With limited time to curate data, we leaned on Generative AI to generate and clean additional datasets.
  • Overly Broad Recommendations 🤔: Our career suggestions felt too generic at first. We refined them by tweaking the scoring system to prioritize hard skills, followed by soft skills, interests, and education.

🏆 What We’re Proud Of

Despite these challenges, we’re thrilled with what we’ve achieved. Hidden Paths offers a data-driven solution to the lack of career guidance, with a Ragnarok Online-inspired graph that makes your career journey easy to visualize. The Q&A feature opens up a world of career exploration, answering a wide range of user questions.


📚 Lessons Learned

Along the way, we learned how to manage data in a Graph Database, leverage ArangoDB for processing, and fine-tune LLMs to generate accurate AQL queries. We also discovered the importance of balancing skill weights in career recommendations.


🔮 What’s Next for Hidden Paths

Looking ahead, we’re excited to take Hidden Paths further. We plan to boost recommendation accuracy by weaving in job market trends and salary data, enhance the Q&A system to tackle more complex questions, and expand to support organizations developing employee skills. With salary analysis and broader insights on the horizon, Hidden Paths is ready to guide even more people toward their dream careers. 🌟

Built With

+ 1 more
Share this project:

Updates