Step 2.5: Project Pitch (30-Second & 2-Minute Spoken Scripts)

30-Second Elevator Pitch

"I built an enterprise-grade omnichannel notification delivery platform using Python, FastAPI, and React that handles mission-critical alerts across Email, SMS, Push, and Webhooks. It features prioritized queues that ensure two-factor authentication codes bypass marketing blasts, automated provider fallback routing that swaps degraded vendors within 50 milliseconds, and an intelligent quiet-hours engine that suppresses non-urgent alerts during recipient nighttime."

---

2-Minute Technical Interview Pitch

"In distributed platforms, notification delivery is inherently brittle because it depends on third-party provider APIs like SendGrid or Twilio that experience intermittent outages, rate limits, and network latency spikes. Furthermore, marketing bulk broadcasts can easily starve urgent transactional OTPs.

To solve this, I designed and built OmniPost, a resilient notification delivery platform.

First, in backend/src/queues/dispatcher.py, the system implements a strict Multi-Tier Priority Queue with four priority tiers: URGENT, HIGH, NORMAL, and LOW. Critical authentication OTPs and security alerts enter with a priority weight of 100 and preempt background marketing newsletters, guaranteeing sub-second dispatch latency regardless of queue depth. To prevent low-priority jobs from starving indefinitely, the queue incorporates aging thresholds.

Second, to insulate the platform against third-party provider outages, I built an automated Provider Selection and Fallback Engine in backend/src/channels/selection.py. When an upstream provider returns HTTP 5xx errors or connection timeouts, the system automatically shifts traffic to a secondary failover provider (such as AWS SES or Vonage) deterministically based on health and cost without dropping messages.

Third, for failed messages, the system implements exponential backoff retries with full jitter in backend/src/retries/failure.py. Messages that exhaust their retry budget are routed to a Dead-Letter Queue with automated manual redrive and replay capabilities, generating new delivery attempts and immutable audit records.

Fourth, an intelligent Preference Routing Engine in backend/src/preferences/routing.py respects user quiet hours based on their local timezone—handling overnight wrap-around windows and Daylight Saving Time adjustments—while enforcing rolling 24-hour frequency capping.

The administrative console provides real-time visibility into queue depth, provider health, and delivery logs. The platform is backed by 114 automated tests running via pytest."