LearnPilot

AI career coach that reverse-engineers job descriptions into personalized learning roadmaps using Gemini 2.5 Flash multimodal vision — turning learners into must-hire candidates

Comment

Inspiration

Career learners are overwhelmed. They Google "how to become a data analyst," find 500 random articles, waste months on wrong tutorials, and still can't get hired. We realized: learning platforms teach skills, but none connect those skills to actual job requirements. LearnPilot bridges that gap using Google's Gemini AI.

What it does

LearnPilot is an AI-powered career roadmap generator built on Gemini 2.5 Flash that solves the "learning-to-hiring" gap:

🗺️ AI Career Architect — Enter your career goal + target market (startup, big tech, government) → Gemini generates a personalized, week-by-week learning roadmap with real resources, capstone projects, and LinkedIn-worthy milestones.

📸 JD Reverse Engineer (Multimodal) — Upload or paste any job description → Gemini analyzes the requirements, identifies YOUR exact skill gaps, and builds a targeted "Must-Hire" roadmap. No other learning platform does this.

🧠 Adaptive Learning — As you complete modules and take quizzes, the AI adapts your path in real-time — harder content if you're excelling, remedial support if needed.

💡 Expert AI Reasoning — Every roadmap includes a "Senior Career Architect" explanation of WHY this path makes you the perfect candidate — not just WHAT to learn, but WHY in that specific order.

How we built it

  • Frontend: React.js with Material UI for a clean, professional Google-inspired interface
  • Backend: Node.js + Express.js with MongoDB Atlas for persistent user data and career paths
  • AI Engine: Gemini 2.5 Flash API for both text generation and multimodal job description analysis
  • Auth: JWT-based authentication for personalized career experiences
  • File Handling: Multer for resume and JD image uploads

The core innovation is our multimodal JD analysis pipeline: users paste or upload job descriptions, Gemini extracts the skill requirements, cross-references them with the user's profile, identifies gaps, and generates a targeted learning path with capstone projects that demonstrate the exact skills employers want.

Challenges we faced

  • AI Response Consistency: Getting Gemini to return consistently structured JSON for complex multi-module learning paths required extensive prompt engineering and robust fallback parsing with regex extraction
  • Model Versioning: Navigating the Gemini model ecosystem during the transition period — older models were being retired while newer ones required different API configurations
  • Schema Flexibility: Designing a career path database schema flexible enough to handle AI-generated content with varying structures while maintaining data integrity

What we learned

  • Gemini 2.5 Flash's multimodal capabilities are incredibly powerful for document analysis — it can extract structured career requirements from messy job descriptions with remarkable accuracy
  • Prompt persona engineering is a game-changer — our "Senior Career Architect" persona produces significantly more actionable and strategic roadmaps than generic instructions
  • Building robust AI integrations requires defensive programming — manual JSON extraction with regex fallbacks is more reliable than depending solely on response format configurations

What's next for LearnPilot

  • 🔗 LinkedIn integration for automated skill gap detection from user profiles
  • 👥 Community-driven path sharing and peer review system
  • 📱 Mobile app for on-the-go learning progress tracking
  • 🎓 Direct integration with course platforms (Coursera, Udemy) for one-click enrollment
  • 📊 Analytics dashboard showing career readiness scores and market demand alignment

Built With

Share this project:

Updates