🔍 Bridging the skill gap with intelligent, personalized course recommendations powered by LLMs, real-world data, and semantic search.
Skill Horizon is an AI-driven platform that leverages real-world job listings and authentic course reviews to identify skill gaps in job seekers and recommend highly personalized online courses. By combining job market analysis, course data scraping, vector-based semantic search, and LLM reasoning, the system delivers recommendations that reflect both industry demand and user-specific learning needs.
Users can enter natural language queries like “real-world projects, hands-on labs, beginner-friendly” to get course suggestions that align with their actual preferences, extracted directly from what real learners have said about the course—making the output more grounded, honest, and actionable.
🚀 Developed as a Hackathon Project at HackHound 3.0
- Analyze a user’s profile (job title, experience, and skills)
- Detect missing skills by scraping and analyzing real-world job listings
- Recommend the most relevant real-world online courses
- Accept natural language input for fine-grained user preferences
- Leverage authentic user reviews to uncover the real value of a course
- Use LLMs and vector databases to deliver the most specific and beneficial courses
- Languages & Tools: Python, MongoDB, Playwright, Scrapy
- AI/ML: LangChain, FAISS, LLMs (Encoders & Decoders), Vector Embeddings
- Web Scraping: Scrapy, Playwright
- Data Storage: MongoDB
- NLP & Retrieval: Semantic search with FAISS + LangChain for vector search and RAG pipeline
-
User inputs:
Job Title,Years of Experience, andCurrent Skills -
Backend process:
- Scrapes live job listings using Playwright & Scrapy
- Extracts required skills from job descriptions
- Compares with user skills to detect personalized skill gaps
- Scrapes real-world courses and reviews from platforms like Coursera, Udemy, and edX
- Courses ranked based on:
- Enrollments
- Ratings
- Authentic learner feedback
- Relevance determined through skill-gap mapping
- User enters personal preferences in natural language (e.g., "real-world projects, labs, beginner-friendly")
- FAISS + LangChain used to semantically search within course descriptions and user reviews
- An LLM analyzes and ranks results to suggest the most relevant, honest, and specific courses that match both the job requirements and the user’s expressed intent
User Input:
- Job: Cyber Security
- Skills: Linux, Network Security
- Experience: 2–4 years
- Query: "real-world projects, lab practicals, and basics covered in detail"
🔍 Output Recommendation:
✅ Recommended Course:
📌 Course Name
🔗 Course Link
📖 Why was this course recommended?
This course covers foundational topics in depth, includes lab-based learning, and matches your preference for real-world, hands-on content—as reflected in user reviews.
- Add resume upload & NLP parsing
- Build user learning paths using curriculum planning
- Enable feedback loop to improve recommendations
- Visual dashboards for in-demand skill tracking
- Harsh Gupta: Developed Generative AI for course selection, semantic search with vector DB, and integrated LLM-based personalization from natural language input.
- Aditya Maurya & Saurabh Tripathi: Scraped real-world job listings and courses; handled data ingestion and MongoDB optimization.
- Shivangi: Crafted and delivered the pitch presentation.
Pull requests and suggestions are welcome! If you'd like to contribute, fork the repo and open a PR.
Harsh Gupta
📧 [email protected]
🔗 LinkedIn | Portfolio
This project is licensed under the MIT License – see the LICENSE file for details.