Pathfinder

Inspiration

PathFinder transforms how individuals upskill by moving away from static, one-size-fits-all course recommendations. It is an Industry-Driven Autonomous Learning Path Generator that cross-references live educational content with real-time, geography-specific job market demands.

Instead of just telling a user how to learn a topic, PathFinder ensures they are learning the exact skills required by employers in their target location, structuring the journey with full prerequisite awareness.

What it does

Traditional learning platforms offer isolated courses without contextualizing them within actual local industry needs (e.g., the specific tech stack demanded by startups in emerging tech hubs).

PathFinder solves these issues through a dual-agent architecture powered by TinyFish and OpenAI, underpinned by a Neo4j Knowledge Graph.

How we built it

The system autonomously researches what the local market needs, finds the best current resources to teach those skills, and maps out the exact sequence in which they must be learned.

Core Workflows

• Geo-Specific Market Signal Extraction Trigger: A user requests a learning path for a specific role and location (e.g., "Full-Stack Developer in India").

Action: TinyFish Web Agents autonomously navigate local job boards and industry forums to extract the most frequently demanded skills, frameworks (like React, Node.js, or Next.js), and tools.

Result: A weighted list of required competencies tailored to the user's geography.

• Curriculum Synthesis & Prerequisite Mapping Trigger: The system identifies a required skill from the market signal that is not yet mapped in the database.

Action: TinyFish agents scrape current educational platforms for top-rated content. The raw, unstructured data is fed into the OpenAI API.

Synthesis: The LLM parses the syllabi, identifies the core concepts, and critically extracts hidden prerequisites (e.g., realizing "Advanced React" requires "Vanilla JavaScript DOM Manipulation").

• Graph Knowledge Retrieval & Storage (Neo4j) The Brain: All synthesized data—skills, courses, geographical demand weights, and prerequisites—is stored as nodes and relationships in a Neo4j Graph Database.

Intelligent Caching: Before dispatching agents for a new query, the system traverses the graph. If a valid, up-to-date path already exists for that location and role, it is retrieved instantly, bypassing the need for expensive API calls.

Evolution: With every new query, the graph grows denser and more interconnected, creating a continuously evolving map of human knowledge and industry demand.

Challenges we ran into

Initially, we fell into the trap of creating generic course recommendation regardless of geography which gave us poor and repeating results. We fixed this by enabling location access to make it more usable.

Use The following email address for access

To run the app use the following email: [email protected]

Built With

• express.js
• neo4j
• node.js
• openai
• react.js