Case Study
Interna Virtual
GDG Hackathon AI internship simulator — 14th of 122 teams, live on GCP.
- Role
- Backend & Database Lead
- Category
- Full-Stack
- Stack
- FastAPI · PostgreSQL · Gemini AI
- Availability
- Live product
GDG Hackathon — ranked 14th of 122 competing teams
The challenge
Built in 72 hours for the GDG (Google Developer Groups) hackathon, Interna Virtual tackles the 'need experience to get a job' paradox. We designed an AI-powered virtual internship where students face real stakeholder friction: AI mentors, code reviews, AI interviewers, adaptive personas, and production-style project workflows — not another tutorial clone.
Architecture
Owned the backend and database layer with FastAPI and PostgreSQL, integrating Google Gemini for mentorship, live code review, and interview simulation. Deployed the full stack on Google Cloud Platform — provisioning services, managing deployments, and operating the console end-to-end. That GCP exposure became one of the biggest technical wins: understanding cloud IAM, service configuration, and shipping under hackathon pressure.
Challenges overcome
Gemini output is non-deterministic — built JSON-schema validation and aggressive prompt engineering so AI responses never broke the Next.js frontend. Parallelized backend API design with frontend integration while configuring GCP services under extreme time pressure.
Lessons learned
Hackathons reward ruthless scope control. The biggest personal takeaway was GCP: going from zero to a deployed cloud stack in 72 hours forced me to learn the platform by doing, not by reading docs in isolation.