Middle Full-Stack engineer test assessment - Description of the BE + FE Application Architecture

1. What We’re Building

• SDK that apps drop in to catch errors and send them to us.
• API that receives those errors and stores them.
• Dashboard where developers search and see the errors.
• Email alerts when a critical error pops up.

---

2. Tech Stack

Part	Tech	Why
SDK	JavaScript snippet published on npm	Works in browser & Node, zero config.
API	Node.js + Express (REST)	Easy to read, tons of examples.
DB	PostgreSQL (JSONB column)	One service, SQL + flexible JSON; cheap on any cloud.
Dashboard	React + Vite	Fast dev server, deploy as static files.
Email	SendGrid API	Quick to set up, free tier.
Deployment	Docker Compose on a small VM (e.g., DigitalOcean)**	One‑command up; can move to Kubernetes later if needed.

---

3. How Data Flows

1. App hits an error → SDK captures stack trace.
2. SDK batches errors and POSTs to /api/logs.
3. API validates request, adds tenant & timestamp, inserts into errors table.
4. API checks: if level = critical, trigger email via SendGrid.
5. Dashboard calls /api/logs?search=… to show data.

---

4. Key Decisions

• REST over GraphQL – fewer moving parts.
• Single DB – avoid extra queues until traffic demands it.
• Batching in SDK – keeps API lightweight without Kafka.
• JSONB – flexible schema, no migrations for every field.
• Docker Compose – easy local dev & prod parity.

---

5. Ops & Monitoring

• Backups – daily pg_dump to cloud storage.
• Uptime – Health‑check endpoint + Pingdom.
• Logs – App logs to file; optional Logtail or Papertrail.

---

6. Questions for the Product Owner

1. Traffic: expected errors per minute now & in 1 year?
2. Retention: how long must we keep logs? 30, 90, 365 days?
3. Auth: do we need per‑user accounts or one team token is fine?
4. Critical definition: what counts as “critical” for alerts?
5. Budget: monthly ceiling for hosting/email?
6. Deployment: any preferred cloud or must be on‑prem?

---

Prepared by Yehor – Full‑Stack JavaScript Developer