Banking UI demo · passkeys · transfers · statements · PWA
UmerPay
Open-source banking UI demo — passkeys, transfers, statements, admin tools. Full-stack Next.js you can clone and run.
Next.js 14App RouterTypeScript
View project →All projects
Gulf lottery · draws, commerce, compliance-aware flows
BIG lottery project in the Gulf
Full-stack lottery platform for the Gulf — tickets, wallets, draws, and compliance-aware journeys. Co-founder & CTO; I lead engineering and delivery end to end.
Production-grade online lottery for the Gulf: many game formats, ticketing, wallets, automated draw lifecycles, and customer-facing results — with operational alerting when publishing paths fail. Positioning is regional Gulf reach, not one emirate or city in isolation.
The stack pairs a modern customer web app with a Django service layer, async workers, real-time channels where it matters, and cloud-backed storage and notifications — sized for Gulf traffic peaks, multi-currency display, and the payment patterns common across the Gulf.
Next.jsTypeScriptReactDjangoDRFPostgreSQLAWS EC2AWS AmplifyRoute 53S3RedisCeleryChannelsMailgunFirebase AnalyticsGitHubPSPs
Highlight
BIG lottery project in the Gulf
Co-founder & CTO — I own technical direction, architecture, and delivery alongside the founding team.
Revenue: Approximate revenue band for the live product: about USD $15K–$20K per month (varies with seasonality, campaigns, and draw cycles — not an audited figure).
Overview
AWS footprint (what runs where)
| AWS service | Role in the product |
|---|---|
| Amazon EC2 | Django REST API, Django admin, Celery workers, supporting services |
| AWS Amplify | Next.js customer app — build, deploy, hosting |
| Amazon Route 53 | DNS, routing, health checks for cutovers |
| Amazon S3 | Static assets, uploads, exports, shared object storage |
Overview
Integrations & tooling (categories)
| Category | How we use it |
|---|---|
| Payments | Multiple PSPs with a provider-agnostic integration layer |
| Transactional email | Mailgun for tickets, auth, alerts, lifecycle mail |
| Product analytics | Firebase Analytics on the web experience |
| Backups & handoff | Database dumps plus Google Drive for ops archive / handoff |
| Source & delivery | GitHub for code, reviews, and release rhythm |
Overview
Core stack (app vs cloud)
| Area | Stack |
|---|---|
| Customer frontend | Next.js (App Router), React, TypeScript, Redux, PWA (Serwist) |
| Backend & admin | Django, DRF, Celery, Redis, Channels |
| Database | PostgreSQL |
| Object storage | S3 (via boto3) — not a substitute for DB backups |
Leadership & delivery focus (illustrative)
How I weighted time as co-founder / CTO — not logged hours.
Backend, draws & reliability35 %
Frontend, UX & performance28 %
Cloud, integrations & vendors22 %
Team, client & roadmap15 %
Revenue band vs reference scale (USD / month)
Bars show the stated band on a 0–25K scale for context only — not audited.
Typical lower month (~floor of band)15 K USD
Typical stronger month (~ceiling of band)20 K USD
Case study
Delivery & product depth
What you can do on the platform (Gulf product)
- Play many game types aimed at Gulf players (quick picks, flagship draws, high-tier draws)
- Use play, cart, account, and live results in one web experience
- Buy and manage tickets with wallet-style balances
- See automated draw schedules, publish windows, and winner-style outcomes
- Read game rules, FAQ, policies, and marketing content in a consistent layout
Tech stack
- Frontend: Next.js (App Router), React, TypeScript, Tailwind-style UI
- Backend: Django for the admin panel and REST APIs (Django REST Framework), JWT-style auth
- State: Redux with persistence; motion with Framer Motion
- PWA: Serwist / service-worker tooling for installable, offline-friendly behavior
- Async work: Celery, Redis, scheduled jobs (django-celery-beat)
- Real-time: Django Channels (WebSockets) where live updates matter
- Data: PostgreSQL; AWS SDK / boto3 for S3 and related object storage
- Transactional email via Mailgun; other channels and programmable voice for urgent ops (e.g. draw not published)
Infrastructure & cloud services
- AWS EC2 — backend deployment (Django API, admin, workers, and related services)
- AWS Amplify — frontend deployment and hosting for the Next.js customer app
- Amazon Route 53 — DNS, routing, and domain management
- Amazon S3 — buckets for static assets, uploads, exports, and shared object storage
- GitHub — source control, branching, reviews, and delivery workflow around releases
- Multiple PSPs (payment service providers) — integrated for checkout and money-movement flows, with abstraction so we are not locked to one vendor
- Firebase Analytics — product analytics, funnels, and behaviour signals on the web experience
- Google Drive — part of the database backup and operational handoff / archive workflow (alongside automated dumps)
- Mailgun — primary provider for transactional email (tickets, auth, alerts, marketing-style drips where applicable)
- Plus logging, monitoring, secrets management, and other integrations as requirements grew — wired pragmatically rather than over-engineered on day one
Challenge — AWS account restriction & same-day migration
- Our AWS environment was suddenly restricted; we had a tight window (on the order of ~6 hours) to restore a working backend and keep dependent services from failing.
- I led moving the workload from the affected AWS account into a new AWS account — not just “spin up a server,” but DNS, data, frontends, and integrations in one coordinated cutover.
- Migrated and re-pointed: EC2 (API, admin, workers), Amazon Route 53 records and hosted zones, S3 bucket data (copy/sync, policies, and access patterns), AWS Amplify for the Next.js frontend, and the surrounding cloud pieces (secrets, env, monitoring hooks) so behaviour matched production expectations.
- After new compute came up, I updated third-party allowlists and IP whitelists — PSPs, webhooks, security appliances, and other vendors — so traffic to the new server and edge addresses was accepted without breaking payments or callbacks.
- Ran focused smoke checks post-cutover (auth, payments, email, draw-related jobs) and stayed on comms with the client and vendors until the riskiest paths were green again.
How draws & operations work
- Draw pipeline: schedule → process → publish, with clear guardrails
- Wheel / reveal-style UX kept in sync with backend state
- Prize distribution and notifications (email, Telegram-style channels)
- Checks and alerts when draws fail to publish or stay unpublished
- Cart, multi-currency display patterns, payments, KYC/AML-oriented account areas
My role & team (how we ship)
- Co-founder & CTO: I lead the product technically end-to-end and still own hands-on full-stack delivery (frontend + backend)
- I mentor 1 junior backend developer and 1 junior frontend developer
- We collaborate with an SEO specialist and a designer
- I manage the client, deadlines, budget, and the delivery pipeline day to day
- I keep priorities clear under pressure and unblock the team
- I take high-level technical and product decisions
- I meet regularly with the client, payment providers (PSPs), and third-party vendors
- I own reliability around draw automation and follow-through when publishing paths fail
Backend and platform architecture
Architecture
Backend & platform
Architecture & integration
- Clear split between the customer Next.js app and Django APIs — transactional REST, WebSockets only where time-sensitive updates matter
- Background workers kept separate from web processes for emails, draw jobs, retries, and housekeeping
- Assets and uploads on S3-compatible storage with sensible permission and signed-URL patterns
- Shared configuration and prize patterns reused across game families to avoid one-off forks
Compliance, content & trust surfaces
- Legal, policy, FAQ, and game-rule pages share a consistent shell and accessible typography — tuned for Gulf-market clarity and repeat use across game families
- Marketing and education content structured so SEO and design can iterate without breaking core flows
- Account journeys include KYC/AML-oriented steps and payment edge cases without naming specific vendors — aligned with how Gulf-facing products typically handle risk and verification
Reliability, monitoring & runbooks
- Management commands and ops hooks for one-off fixes, backfills, and reporting
- Guardrails on draw publication and “still unpublished” detection with escalation paths
- Programmable voice (and other channels) for highest-severity failures such as missed publishes
- Defensive patterns on heavy tasks — memory awareness, circuit-breaker style backoff, and idempotent jobs
Stakeholders, cadence & leadership
- Regular design and SEO syncs so launches stay on-brand and discoverable
- Pairing and structured code review with junior engineers; clear Definition of Done before merge
- Working sessions with the client and PSPs on payment behaviour, timelines, and trade-offs
- Technical calls with third-party vendors for integrations, SLAs, and cutover windows
Interested in similar delivery?
Get in touchKeep exploring
Related projects
More work with a similar stack and delivery focus.
Food delivery · React · NestJS · admin
Ghazi Restaurant
Online food ordering & restaurant management — React customer app, NestJS APIs, MongoDB, admin dashboards, chat, and maps. Technical Lead (40% coding, 60% leadership).
ReactNestJSTypeScript
View project →Morocco · AI matchmaking · 100K users
Cuddloo - AI Match Making App
AI matchmaking and dating app for Morocco — Tinder/Bumble-style product with 100K users. I led Android (Kotlin), Node.js backend, PostgreSQL, web, iOS supervision, and AWS delivery.
KotlinAndroidSwift
View project →© 2026 Umer Sheraz