Inspiration

The modern job market is a moving target. Professionals are stuck in a loop of manual job hunting and "tutorial hell," learning skills that might be obsolete by the time they finish. We wanted to build a system that doesn't just list skills, but acts as a living GPS for your career—finding the path, adjusting for traffic (market shifts), and proving you’ve actually arrived (verification).

What it does

Skill Horizon is an autonomous career intelligence system. It uses an AI agent to scout LinkedIn and Indeed based on your "Career Vibe," extracts the most in-demand skills, and builds a dynamic "Upskill Mission." Unlike other platforms, it uses Gemini Live to conduct real-time voice interviews and Gemini Vision to watch you code, verifying your mastery so you have proof for employers, not just a checkbox.

How we built it

The Brain: Google Gemini 2.0 Flash for high-speed skill extraction and market trend analysis.

The Voice & Eyes: Gemini Live API for multimodal voice assessments and Gemini Vision for real-time code/work review.

The Agent: A Playwright-based automation engine that navigates job boards to scout opportunities autonomously.

The Stack: Built with React, TypeScript, and Tailwind CSS, powered by a Node.js backend with PostgreSQL for tracking long-term skill "health" and mission progress.

Challenges we ran into

Integrating Gemini Live for real-time conversation was a hurdle—balancing latency so the AI interviewer felt human and not laggy. We also had to solve the "Scraper Cat-and-Mouse" game, implementing rotating proxies and human-like browsing patterns to ensure our autonomous scout could find jobs without getting blocked by anti-bot measures.

Accomplishments that we're proud of

We successfully built a Market Shift Detection algorithm that actually works. Seeing the system automatically "promote" a skill like Next.js in the user's roadmap because it detected a spike in job postings felt like watching a real career coach at work. We also created a seamless screen-share integration that provides live "mentor-style" feedback while a user codes.

What we learned

We learned that AI shouldn't just be a tool; it should be an agent. Moving from a "search-based" UI to an "intent-based" UI (The Career Vibe) completely changed how users interacted with the platform. We also mastered the intricacies of WebRTC and multimodal streaming to make the Gemini-led assessments feel truly interactive.

What's next for SKILL HORIZON

The next step is "Horizon Portfolios"—shareable, cryptographic proof of skill mastery that users can embed directly into their LinkedIn profiles. We also plan to integrate with Google Calendar to auto-schedule "Interview Prep" sessions with the AI when a user hasn't practiced a critical skill in a while.

Built With

Share this project:

Updates