AI EngineeringPrivate
AI University
A full-stack learning platform for Logius, built as a school assignment with my project group, featuring an LLM-powered AI Coach.
Full-stack developer within the project group for Logius · 2025 — 2026
- React
- Next.js
- TypeScript
- Tailwind CSS
- .NET
- C#
- Docker
- PostgreSQL
Built: prompt → LLM → chat response
View source code on GitHub ↗
Problem
Professionals need guidance on using generative AI safely and practically within a learning platform.
Approach
An AI Coach integrated as a chat feature in AI University, giving users immediate answers alongside modules, cases and exercises.
Engineering decisions
- Designing the AI Coach as part of the existing platform rather than a standalone demo
- Placing the LLM call behind the application layer to keep the chat flow and integration manageable
- Combining AI functionality with the existing frontend/backend separation and authentication
Read the technical details
AI University is a school assignment for Logius, a government organization. Our project group was the only group working on this assignment for Logius. The platform helps government professionals use AI safely, critically and effectively. Within the group project, I integrated an AI Coach alongside the learning path, practical cases and gamification. The user sends a message, the application builds a prompt from it, sends it to an LLM and displays the response in the chat. This is a concrete LLM integration within a larger platform, not a RAG or agent system. The Next.js frontend and ASP.NET Core backend are separate, with Nginx as a reverse proxy and Docker Compose for local development.
AI component
The user sends a message, the application creates or uses a prompt, sends it to an LLM and displays the response in the chat. The value lies in the learning platform integration and user experience, rather than advanced retrieval or agent orchestration.
Features
- A Duolingo-style learning path with four AI literacy levels
- Interactive games: Red vs Blue, AI Detector, Prompt Duel
- Badges, XP, levels and streaks for user motivation
- Practical cases and coaching for everyday AI use
- JWT authentication with edge proxy and API-level authorization
- Docker Compose for local full-stack development
Challenges
- Building a complete AI literacy platform with separate frontend and backend architectures
- Seamlessly integrating gamification (badges, XP, levels and streaks) with learning content
- JWT authentication and authorization with edge-level proxy checks and backend validation
- Local development with Docker Compose while using Render for hosted deployment
Lessons learned
- How a Next.js frontend and ASP.NET Core backend work together through an Nginx reverse proxy
- Gamification design — how badges, XP and levels strengthen user motivation
- Working with JWT authentication in a full-stack architecture
- Docker Compose for reproducible local development with multiple services