Services and systems I've architected and shipped — a Kafka notification pipeline, payment cores, an AI microservice, and the infrastructure under them. Each card leads with the system I owned and the guarantee it makes.
A food-delivery platform that needs reliable, production-grade payment collection from day one.
Role · Took a NestJS food-delivery backend from code to live production solo — owned the AWS deploy, reverse proxy, DNS, and security hardening.
AWS EC2 + Dockerized stack behind an Nginx reverse proxy and Cloudflare DNS, hardened with fail2ban + a WAF. Paystack payments live in production, collected and verified end-to-end.
A multi-sided food-delivery marketplace needs three tightly-coupled mobile clients — consumer, vendor, and rider — kept in lockstep on a single release cadence.
Role · Built the NestJS API and the transactional-outbox notification pipeline behind the three clients.
Exactly-once push delivery across FCM and APNs via an outbox + dedup key, even under producer retries.
Sending money on Telegram without exchanging account details — peer-to-peer transfers that work even when two phones share only the air between them.
Role · Architected and built the system end-to-end: ultrasonic token transmission, payments core, bots, and infrastructure.
HMAC-SHA256-signed payment tokens transmitted over ultrasonic sound — no internet on the receiver — settled under Serializable isolation with an idempotent ledger, deployed as a full Docker Compose stack on a VPS.
Informal traders in Nigeria run their businesses on WhatsApp and feature phones — they need an AI assistant that meets them on WhatsApp, USSD, and voice.
Role · Owned the full technical architecture and build: a Go comms microservice fronting a NestJS core, with Gemini Live powering voice.
Multi-channel AI assistant (WhatsApp / USSD / voice) for informal traders — a Go comms microservice split from a NestJS core, with Google Gemini for language. Built for the OPay × Google National Innovation Challenge 2026.
Informal construction workers have real, verifiable labor but no credit history a bank can read — so they stay locked out of financing.
Role · Wired the payments and identity backend — Squad APIs for disbursement and Smile ID KYC — behind the worker app and dashboards.
A verified-labor → credit-history pipeline with real payment + KYC integrations, built solo for Squad Hackathon 3.0.
The reliability patterns behind production infrastructure — exactly-once delivery, backpressure, retries — are easy to name and hard to get right, so I am building them from the ground up.
Role · Currently building — Designing and building it solo as a deep-dive into reliable infra: a Kafka broker, Redis rate-limiting, Postgres persistence, and channel workers with retry/backoff and dead-letter queues.
Building a distributed notification system in Java/Spring Boot: Kafka broker, Redis rate-limiting, Postgres persistence, channel workers with retry/backoff, dead-letter queues, and idempotent delivery. A ground-up study in the systems patterns behind reliable infrastructure.
Multiple hackathon wins — I build fast, ship working systems under deadline, and present them.
OPay × Google National Innovation Challenge 2026
Pitched and prototyped against a national field — an AI assistant that meets informal traders on the channels they already use.
SwiftyEx Hackfest
Telegram P2P payments over ultrasonic sound with signed, idempotent tokens.
Squad Hackathon 3.0
Construction supply-chain fintech connecting builders with material financing.
Bayse Hackathon
AI portfolio-risk tool surfacing exposure and synthetic-asset risk signals.
The design decisions behind these systems — correctness under failure, and scale.
SwiftHum's ledger uses HMAC-signed tokens with Serializable isolation so a replayed or duplicated transfer can never double-spend — the same discipline as the notification service's dedup layer.
OWise splits a Go comms microservice from the NestJS core so the latency-sensitive channel layer scales independently of business logic.
The notification system applies the patterns that keep infra up: retry with exponential backoff, dead-letter queues, and at-least-once delivery made safe by idempotency.
Faith Popoola — backend & distributed-systems engineer, Founder / Backend Engineer at QuickBite since April 2026.
I'm drawn to systems that have to stay correct under pressure — payment ledgers, message queues, idempotent delivery. I deploy and harden my own infrastructure (AWS, Hetzner, Docker, Nginx, Cloudflare) and I'm going deep on distributed systems through a Java/Spring Boot notification service built on Kafka, Redis, and Postgres.