SkillLink AI

Millions of rural youth in India possess untapped skills but lack access to opportunities, guidance, and digital tools to succeed. Inspired by conversations with local NGO workers and observing skill gaps in rural communities, we realized that voice-first, AI-powered guidance could bridge this gap — especially for non-English speakers. We wanted to build something inclusive, intuitive, and startup-scalable — a platform that could listen, learn, guide, and connect. That became SkillLink AI.

How to build voice-enabled GPT workflows with Whisper and GPT-4o

The importance of local language UX in rural tech adoption

Building for low-bandwidth, low-literacy environments

How to design human-centric AI flows that feel like natural mentorship We started by designing the voice conversation flow, then trained prompts for GPT-4o to handle skills assessment, goal setting, and job matching.

Using Whisper API, we converted voice input to text in regional languages.

For the backend, we used Firebase and Flask, with RapidAPI for scraping nearby job opportunities.

The frontend is built in Flutter for mobile (Android-first), with a fallback React web app.

We localized everything — content, feedback, and prompts — to support Telugu, Hindi, and voice-only navigation. Challenges We Faced Handling multilingual voice inputs without errors

Designing GPT prompts that sound natural and empathetic to first-time users

Matching real jobs based on partial skill info

Building an accessible UX for low-literacy users

Keeping things functional on 2G/3G networks

Built With

• and-job-matching.-using-whisper-api
• and-prompts-?-to-support-telugu
• dart-frameworks:-flask
• feedback
• firebase
• flutter-ai-tools:-openai-gpt-4o
• goal-setting
• google-maps
• google-translate-cloud:-render
• hindi
• javascript
• langchain-databases:-firebase-firestore-apis:-rapidapi-(job-search)
• languages:-python
• react
• then-trained-prompts-for-gpt-4o-to-handle-skills-assessment
• vercel
• we-converted-voice-input-to-text-in-regional-languages.-for-the-backend
• we-used-firebase-and-flask
• whisper-api
• with-a-fallback-react-web-app.-we-localized-everything-?-content
• with-rapidapi-for-scraping-nearby-job-opportunities.-the-frontend-is-built-in-flutter-for-mobile-(android-first)