SparkPath

Inspiration

Youth poverty is a persistent crisis, and we noticed a critical gap: young people dream of careers in entertainment but lack access to guidance on how to get there. Traditional career counseling is expensive, generic, and disconnected from creative industries. We were inspired by the idea that AI could democratize career discovery—providing personalized, instant guidance that's usually only available to the privileged. SparkPath was born from our belief that everyone deserves a clear path to their dream career, regardless of their background.

What it does

SparkPath is an AI-powered career discovery platform that guides underserved youth into entertainment careers through four key features: AI Career Assessment - A conversational chatbot asks 4 targeted questions to understand interests and recommends the perfect career path from 41+ entertainment careers across 6 categories (Film, Music, Sports, Animation, Writing, Business). Personalized Dashboard - Users receive a custom 5-step career pathway generated by AI, success stories matched to their demographics (race, ethnicity, location), and mentor recommendations from professionals who share their background. Adaptive Learning - Course catalog with real-time progress tracking. At 25% completion, an AI wellness check intervenes to ensure they're happy with their path—reducing dropout rates. Networking & Certification - Connect with peers who completed similar courses, earn certificates upon completion, and access job opportunities—all in one platform.

How we built it

Node.js + Express - RESTful API with Socket.io for real-time chat AWS Bedrock - Claude 3 Sonnet for AI career analysis and recommendations AWS DynamoDB - 9 NoSQL tables with Global Secondary Indexes for flexible querying JWT + bcrypt - Secure authentication Frontend: React 18 + TypeScript - Component-based UI with type safety Vite - Lightning-fast development and builds Tailwind CSS + shadcn/ui - Beautiful, accessible UI components Socket.io-client - Real-time bidirectional communication Architecture: Service layer pattern separating business logic from data access Predetermined questions for instant chatbot responses, with AI analysis only at the end Composite primary keys in DynamoDB for efficient relationship modeling ES modules throughout for modern JavaScript Deployment: AWS EC2 with PM2 process manager and Nginx reverse proxy

Challenges we ran into

AI Performance - Initially, we used Bedrock to generate every chatbot question, causing 3-5 second delays. Users felt frustrated. We pivoted to predetermined questions with AI analysis only at the end, reducing response time to instant while maintaining intelligent recommendations. 2. DynamoDB Schema Design - Coming from relational databases, designing for NoSQL was challenging. We learned to use composite primary keys (userId#courseId) and GSIs strategically for access patterns like "find all courses by category" or "find mentors by location." 3. Real-time State Management - Keeping WebSocket sessions synchronized with database state was tricky. We implemented in-memory session maps for active conversations while persisting everything to DynamoDB for history. 4. Demographic Matching Algorithm - Balancing AI recommendations with demographic filters was complex. We needed success stories and mentors to be both relevant to career interests AND representative of the user's background. 5. Wellness Check Logic - Determining when to trigger adaptive interventions and how to analyze sentiment from open-ended responses required careful prompt engineering with Bedrock.

Accomplishments that we're proud of

✅ Complete End-to-End Journey - From signup to assessment to learning to certification to networking—we built the entire pipeline, not just a proof of concept. ✅ Strategic AI Integration - We optimized Bedrock usage to be instant where it matters (chatbot UX) and intelligent where it counts (career analysis). 4 questions in, AI makes accurate recommendations across 41 careers. ✅ Production-Ready Architecture - 9 properly designed DynamoDB tables, service layer abstraction, error handling, authentication, and deployment configuration. This isn't just a hackathon demo—it could scale. ✅ Inclusive by Design - Demographic matching for success stories and mentors ensures underrepresented youth see role models who look like them—representation matters. ✅ Adaptive Learning - The 25% wellness check is innovation in EdTech. Most platforms let students silently drop out. We intervene early and offer alternatives. ✅ Beautiful, Functional UI - With shadcn/ui and Tailwind, we built a professional, accessible interface that looks like a real product.

What we learned

Technical Skills: AWS Bedrock prompt engineering for structured outputs NoSQL schema design with DynamoDB composite keys and GSIs Real-time architecture with Socket.io and session management Modern React patterns with TypeScript and context Strategic AI usage—knowing when NOT to use AI is as important as knowing when to use it Product Insights: Speed matters more than perfection in chatbots—predetermined questions feel better than slow AI generation Representation matters—demographic matching creates emotional connection Early intervention (25% wellness check) prevents dropouts better than late-stage recovery Complete journeys convert better than partial solutions Teamwork: Iterating quickly based on UX feedback Balancing ambitious vision with hackathon time constraints Documentation matters—our 8 markdown files make the project accessible to judges and future contributors

What's next for SparkPath

Short-term (Next 3 months): Pilot Program - Partner with 2-3 youth organizations to test with real users (50-100 students) Content Expansion - Add 50+ more courses across all 41 career subcategories Mobile App - React Native version for iOS/Android accessibility Analytics Dashboard - Track completion rates, popular careers, demographic insights Medium-term (6-12 months): Live Mentorship - Video call integration with Zoom/Google Meet APIs for scheduled mentorship sessions Job Board - Partner with entertainment companies for entry-level job postings Community Features - Discussion forums, peer study groups, success story submissions Internationalization - Multi-language support (Spanish first) Long-term Vision: Scale Beyond Entertainment - Expand to STEM, Healthcare, Education, Trades Partnership Network - Integrate with industry organizations (SAG-AFTRA, Recording Academy, etc.) Financial Aid Integration - Connect users to scholarships, grants, equipment loans Impact Measurement - Longitudinal studies tracking career outcomes and income mobility Monetization Strategy: Free for students (always) B2B partnerships with youth organizations ($50-100/student/year) Premium features for career advisors (analytics, bulk user management) Employer partnerships for talent pipeline access Our Mission: Reduce youth poverty by 50% by making career pathways accessible to everyone, starting with entertainment and expanding to all industries. SparkPath is just the beginning.