RiseMap

Inspiration

Navigating career choices can be overwhelming. Most tools today are outdated, static, or rely heavily on keywords. We were inspired to build an intelligent platform that feels like a career coach—offering real-time, tailored guidance using AI, web search, and curated data. Thus, we would like to introduce RiseMap.

RiseMap is an intelligent career discovery platform that helps users find their perfect career path using semantic search and AI assistance. The system uses machine learning embeddings to match user queries with relevant career opportunities from the O*NET (the U.S. Department of Labor's primary source of occupational information) database.

What it does

RiseMap helps users explore and evaluate careers through natural language queries. It semantically matches user input to career profiles from O*NET - the U.S. Department of Labor database and enriches results with real-time data like salaries, hiring trends, and work culture using the Tavily API. Users can chat with an AI assistant through CopilotKit, compare paths, and plan smarter.

🔍 Semantic Career Search

• Natural language career queries (e.g., "software development", "helping people")
• AI-powered similarity matching using sentence transformers
• Real-time search results with relevance scoring

🤖 CopilotKit AI Assistant Integration

• CopilotKit-powered chat interface
• Natural language career search commands
• Contextual career guidance and recommendations

📊 Tavily Career Enrichment

• Real-time salary information from Glassdoor
• Work culture insights from Reddit discussions
• Top hiring companies from Indeed
• Expandable/collapsible detailed information

🎯 Advanced Matching

• HuggingFace sentence-transformers model (all-MiniLM-L6-v2)
• Cosine similarity scoring for career relevance
• O*NET SOC code integration for standardized career classification

How we built it

Machine Learning

• HuggingFace Transformers - Sentence embeddings generation
• MLflow - Model deployment and experiment tracking
• scikit-learn - Similarity calculations and ML utilities
• PyTorch - Deep learning framework

Backend

• Flask - Web framework with CORS support
• Tavily API - Real-time web search for career enrichment
• Pandas - Data processing and manipulation
• Python-dotenv - Environment variable management

Frontend

• Next.js 14 - React framework with TypeScript
• CopilotKit - AI assistant integration
• Tailwind CSS - Utility-first styling
• Framer Motion - Animation library

Challenges we ran into

• Integrating CopilotKit with dynamic in-page actions
• Handling rate limits and formatting for Tavily’s real-time search
• Ensuring semantic search accuracy across diverse queries
• Maintaining responsiveness across API layers (Flask, MLflow, Tavily, frontend)
• I had trouble deploying the backend code in production. My backend and frontend works well and functions as expected during local test.

Accomplishments that we're proud of

• Seamlessly combined AI, real-time data, and semantic search
• Built a fully working E2E system with real-world job data
• Designed a smooth user experience using CopilotKit, Tavily, and interactive filters
• Achieved sub-second response time for 1000+ career embeddings

What we learned

• How to embed and deploy NLP models using MLflow
• Building full-stack apps with AI-driven backends and Next.js frontends
• Designing AI assistants (CopilotKit) that go beyond static answers
• How to enrich structured data with live, unstructured web content using Tavily

What's next for RiseMap

• Add personalized recommendations based on user goals and past experience
• Integrate mentorship scheduling and resume-building tools
• Expand career data to include freelance/gig economy roles
• Launch a mobile-friendly version and gather user feedback for version two

Built With

• copilotkit
• flask
• framer-motion
• huggingface
• mlflow
• next.js
• pandas
• python
• pytorch
• scikit-learn
• tailwind
• tavily