Team 79_SkillTree

Inspiration

The inspiration for SkillTree stems from the rapid evolution of Singapore’s data-driven economy. We noticed a significant information disconnect: while the job market moves at light speed, the flow of actionable data between employers and talent remains fragmented. Traditional job searching relies on static job titles and keywords, which fails to capture the true potential of individuals or the specific, evolving needs of companies. We built SkillTree to be the bridge that translates market data into a shared, dynamic language of skills.

What it does

SkillTree is a career intelligence platform that shifts the focus from "what your title is" to "what you can do." It aggregates real-time market data to provide:

Automated Skill Profiling: Instantly parses resumes and LinkedIn profiles using high-fidelity AI to build a comprehensive digital skill map.
Dynamic Suitability Scoring: Employs a graph-based engine to show users exactly how they stack up against specific roles with a real-time "Suitability Score."
Visual Gap Analysis: Interactive Knowledge Graphs visualize the "delta" between a user's current competencies and a job's specific requirements.
Targeted Upskilling & AI Roadmaps: Instead of a "No," the platform provides "Not Yet" by offering curated roadmaps, learning resources, and MCQ assessments to close specific skill gaps.
Real-time Interview Coaching: A voice-activated AI mentor ("Final Boss") that conducts realistic mock interviews and provides instant, actionable feedback on performance.
Ubiquitous Integration: A dedicated Chrome extension that brings SkillTree’s intelligence directly onto LinkedIn and other job portals.

How we built it

We developed a full-stack ecosystem using a modern, high-performance tech stack:

Backend: Built with FastAPI for high-concurrency performance and Supabase for robust database management. We integrated Google GenAI (Gemini 2.0 Flash) for semantic understanding and NetworkX to manage complex graph relationships between competencies.
Frontend: A premium interface built on Next.js 16 (React 18.3) and Tailwind CSS. We used ReactFlow and Recharts to turn abstract data into digestible, interactive visualizations.
Real-time Interaction: Leveraged Pipecat AI to power our voice-driven interview coaching, integrated with Deepgram (STT) and Cartesia (TTS) for lifelike, low-latency conversation.
Data Pipeline: Utilized PyMuPDF and Pandas for robust parsing and processing of resumes and large-scale job market datasets.