How learnathon-by-geeky-solutions/earendel works
Project Type & Purpose
• This is a full-stack Interview as a Service platform, with a modular monolithic .NET 8 backend exposing APIs for multiple Angular front-ends (Interviewer, Candidate, Admin, HR).
• It uses Clean Architecture, Domain-Driven Design (DDD), CQRS, Event-Driven patterns and supports multi-tenancy.
Key Architectural Layers & Components
A. External Clients
• Angular UIs (Interviewer, Candidate, Admin, HR portals)
• Swagger / API testing tools
B. Edge & API Gateway
• NGINX (ingress)
• Aspire API & Service Management
C. Backend Host
• ASP.NET Core Web Host (Api/Server and Aspire.AppHost)
• Framework Core & Infrastructure libraries (cross-cutting concerns: Auth/JWT, Audit, Caching, Logging, HealthChecks, Rate-Limiting, SignalR, Hangfire)
• Shared library (Authorization, Constants, Extensions)
D. Modules (each follows DDD + Clean Arch)
• Candidate, Jobs, Evaluator, Experties, HRView, InterviewerView, Interviews, Notifications, Quizzes – each with
– Domain (entities, events, exceptions)
– Application (commands, queries, handlers, validators)
– Infrastructure (EF Core DbContext, Repositories, Endpoints, Messaging consumers)
E. Data & Persistence
• PostgreSQL (per-tenant or shared schemas)
• EF Core migrations & DbInitializers
F. Messaging & Real-Time
• RabbitMQ (domain events, inter-module messaging)
• SignalR (real-time notifications)
G. Background Processing
• Hangfire (scheduled & fire-and-forget jobs)
H. Caching & External Services
• Redis (distributed caching)
• SMTP / Mail service
• Zoom SDK integration
I. Monitoring & Observability
• Prometheus (metrics scraping)
• Grafana (dashboards)
• Aspire / Serilog (logging)
J. Deployment & Infra
• Docker & docker-compose
• Kubernetes manifests (deployments, services, persistent volumes)
Relationships & Data Flows
• UI → HTTP/HTTPS calls → NGINX → ASP.NET Web Host → Module Endpoints →
– Commands → Mediator → Application Handlers → Domain & Infrastructure → PostgreSQL
– Queries → read models → PostgreSQL
• Application Handlers publish DomainEvents → RabbitMQ → subscribed Module Consumers
• Background jobs (Hangfire) enqueue & process tasks via the same host
• Caching layer (Redis) wraps read/write for high-speed lookups
• SignalR pushes notifications (e.g., new assignments, interview status) back to UIs
• Host exports metrics (Prometheus endpoints) → Prometheus scrapes → Grafana visualizes
• Aspire monitors API traffic & service health; logs shipped via Serilog to Aspire
Architectural Patterns & Principles
• Modular Monolith: modules loaded into one process, clear module boundaries
• Clean Architecture: separation into Core, Application, Infrastructure, Presentation
• DDD: Entities, Value Objects, Domain Events per module
• CQRS: distinct command and query models
• Event-Driven: asynchronous messaging via RabbitMQ
• SOLID, Dependency Injection, Multi-Tenancy via tenant middlewares