SkillSync AI

Inspiration

Every year in Kenya, thousands of university students dedicate years to earning academic degrees, only to graduate into a frustrating paradox: they send out countless CVs without response, while employers simultaneously report a significant talent shortage. We recognized that the issue is not a lack of talent, but a measurable skills gap between traditional academic training and rapidly evolving industry demands. Students often guess which skills matter, pursue misaligned courses, and miss career-defining opportunities. We set out to build a bridge — a system that transforms job hunting from uncertainty into a precise, data-driven process. That vision became SkillSync AI.

What It Does

SkillSync AI is a Career Intelligence Copilot designed for the Kenyan job market. Users paste their CV or list of skills into our intuitive interface and select a target career path (e.g., Software Engineer, Financial Analyst, Clinical Officer). Our platform then performs a deep AI-powered semantic alignment against curated Kenyan job market data and returns: Alignment Score – A percentage (0–100%) showing how job-ready the candidate is. Skill Validation & Gap Analysis – Identifies strengths and clearly outlines missing modern skills. Actionable Sprints – Suggested projects and relevant certifications (local and global). 30-Day Roadmap – A structured week-by-week plan guiding the user toward job readiness in four weeks.

How We Built It

Frontend Ecosystem We built the interface using React and Vite, styled with Tailwind CSS. We focused on a premium, glass-morphic dark aesthetic with full mobile responsiveness. Backend Architecture The backend runs on a Node.js Express API handling skill parsing and AI orchestration. AI Engine We integrated Google’s Gemini 2.5 Flash API. To ensure reliability, we used Structured Output (JSON Schema), forcing the model to return predictable, deeply nested JSON responses. This guarantees UI stability and eliminates unpredictable formatting issues.

Deployment

The project is structured as a monorepo and deployed on Vercel’s Edge Network. Custom rewrite rules manage frontend static assets and serverless backend functions efficiently.

Challenges We Ran Into

Our biggest challenge was resolving compatibility issues between modern ES Modules and legacy CommonJS in a serverless environment.

Vercel initially produced 500 Internal Server Errors due to module resolution conflicts between the Vite frontend and backend dependencies.

We solved this by:

Refactoring into a clear frontend/ and backend/ monorepo structure

Separating compilation tracks

Rewriting deployment targets in vercel.json

Ensuring clean module isolation between environments

This stabilized the deployment pipeline and eliminated runtime crashes.

What We Learned

We learned the true power of Structured AI Outputs. By enforcing strict JSON schemas, we transformed an LLM from a simple text generator into a deterministic functional component within our software architecture.

We also gained deep experience in:

Monorepo architecture

Serverless edge deployment

Routing optimization in Vercel

Full-stack system integration

What’s Next for SkillSync AI

Our next steps include:

Expanding the Kenyan job market dataset to thousands of dynamic roles across Africa

Introducing LinkedIn integration to automate profile ingestion

Building an employer portal where recruiters can discover candidates aligned with specific hiring roadmaps

Adding real-time labor market analytics for adaptive skill recommendations

Our long-term vision is to make career alignment measurable, actionable, and accessible across emerging markets.

Built With

• css3
• genai
• google
• html5
• javascript
• lucide-icons
• react
• shadcn-ui
• space-grotesk-&-inter-fonts-backend-&-routing:-express.js
• styling-&-ui:-tailwind-css
• typescript