Question Bank Platform — a 23-project .NET solution, shipped solo
End-to-end quiz and content management platform. 23-project layered .NET 8 solution, three audience-specific APIs (student / admin / teacher), Angular admin panel, AWS S3 media pipeline, Redis caching, Hangfire jobs — serving 10K+ students.
- Students
- 10K+
- Projects in solution
- 23
- Audience APIs
- 3
- Ownership
- Solo
Tech stack
- .NET 8
- ASP.NET Core Web API
- EF Core
- SQL Server
- Angular
- TypeScript
- AWS S3
- AWS MediaConvert
- Redis
- Hangfire
- Firebase
- Docker
Situation
The education platform needed a dedicated Question Bank system: teachers author questions, an admin team curates and tags them, students consume them through quizzes embedded in the main app. Existing tooling was a Google-Sheets-plus-prayer workflow that didn’t scale and blocked content growth.
Task
Build the Question Bank platform end-to-end, as the sole developer: data model, three audience-specific APIs, Angular admin panel, background processing, cloud storage pipeline, AI-assisted content review. Ship it fast enough to unblock content ops, solid enough to survive 10K+ student load.
Action
Solution architecture. I designed a 23-project .NET 8 solution along Clean Architecture lines — Domain / Application / Infrastructure / Persistence / three API hosts / shared contracts / testing. Each audience (student / admin / teacher) gets its own API surface so authorization, DTOs, and rate limits can diverge without cross-cutting hacks.
- Student API — read-optimized, heavily cached in Redis, signed URLs for media
- Admin API — CRUD, bulk operations, content review workflows, full audit trail
- Teacher API — authoring flows, draft/publish states, ownership scopes
Angular frontend. Admin panel built in Angular + TypeScript: question authoring, tagging taxonomy, media attachment, bulk import/export, review workflow, analytics. Component-driven, lazy-loaded modules, RxJS for async flows.
Media pipeline. Questions often include images and audio. I wired AWS S3 for storage with MediaConvert for audio normalization, and signed URLs so the student API never leaks direct bucket access.
Background processing. Hangfire for scheduled jobs (content re-indexing, usage rollups, cleanup), Redis for caching hot question sets, Firebase for push notifications on workflow events (e.g., “your question was approved”).
AI-assisted content review. Integrated an AI pipeline that pre-screens newly authored questions for quality issues (unclear phrasing, duplicate detection, difficulty estimation). Human reviewers now focus on edge cases instead of routine triage.
DevOps. Dockerized, deployed to Azure, CI/CD via GitHub Actions. Swagger/OpenAPI on every API for the frontend and QA teams.
Result
- 10K+ students served through the platform
- Admin content throughput went from spreadsheet-bound to parallel multi-reviewer workflows
- AI review pipeline reduced manual administrative effort significantly — reviewers see fewer obviously-flagged items
- Zero cross-audience security incidents thanks to the per-audience API split
- Now serves as the content backbone for quizzes embedded across the broader ed-tech platform
Why this project matters for clients
This is the case study to read if you need someone who can own an entire product slice. Architecture, data modeling, multiple API surfaces, Angular admin, cloud integrations, background workers, AI hooks — all shipped by one engineer, with the code quality to onboard others later.