PathForge — Career Intelligence Platform
PathForge — Career Intelligence Platform
Overview
PathForge democratizes enterprise-grade career intelligence — Skills Graphs, Market Signals, and Predictive Analytics — for every professional, not just those inside large corporations. It is a full-stack, AI-native product: a Next.js 15 + TailwindCSS v4 web experience backed by a FastAPI (Python 3.12) service that orchestrates LLM pipelines over Claude Sonnet 4 and Gemini Flash, with Voyage AI v3 embeddings stored in PostgreSQL 16 + pgvector for semantic matching. Redis handles JWT blacklisting and an ARQ task queue, authentication is stateless via access/refresh JWTs, and the whole system ships through GitHub Actions quality gates to Vercel (web) and Railway (API). The result is a career platform where opportunities are built around the person — 'not just another job board.'
Semantic Matching (pgvector + Voyage AI)
LLM Pipelines (Claude Sonnet 4 · Gemini Flash)
Next.js 15 + FastAPI Full-Stack
CI/CD to Vercel + Railway
Challenge & Solution
The Challenge
Enterprise career intelligence — skills graphs, market signals, predictive analytics — has always been locked inside large organizations, while everyone else is left with keyword-matching job boards that ignore trajectory and ambition.
The Solution
Built an AI-native platform that models each professional's 'Career DNA' and matches it against live market signals using vector search. A FastAPI backend orchestrates multi-model LLM pipelines and Voyage AI embeddings over pgvector, while a Next.js 15 frontend delivers the experience — all deployed through automated CI/CD quality gates.
Key Features
Career DNA™ Engine
Models skills, trajectory, and ambitions as a semantic profile, then matches it to opportunities via pgvector similarity search.
Multi-Model AI Pipelines
FastAPI orchestrates Claude Sonnet 4 and Gemini Flash with Voyage AI v3 embeddings for analysis, matching, and personalized strategy.
Production Architecture
Next.js 15 + TailwindCSS v4 frontend, PostgreSQL 16 + pgvector, Redis queue/cache, and stateless JWT auth.
Automated Delivery
GitHub Actions quality gates ship the web app to Vercel and the API to Railway on every merge.