Selected work

VIA

Building a multi-product marketplace and logistics platform.

Role
Chief Technology Officer
Period
2025 — Present
Context
Early-stage marketplace and logistics platform, Egypt
Scope
Product, architecture, backend, web, mobile, infrastructure, security, production operations, governance, hiring
VIA's website: “Groceries that move at your pace”, with checkout, rider-arrival and store cards around the headline.

An operational problem, nine product surfaces.

VIA is an early-stage multi-sided marketplace and logistics platform for the Egyptian market. I lead product and engineering across nine product surfaces — translating marketplace, logistics and operations needs into workflows, architecture and shipped web and mobile products.

The platform today: seventeen application services, a web platform, six applications, cloud infrastructure and production observability. What follows is how it is run, not a map of how it is wired.

From two services to seventeen.

Led the evolution from a two-service backend into a seventeen-service, database-per-service platform: sixteen domain services behind a single gateway. I created eleven of those services, across orders, payments, ledger, inventory, dispatch, notifications and internal operations.

Boundaries follow domains, not technologies. Complexity is allowed to exist — behind clear edges, with one public surface.

What the system can do.

  • Logistics & geospatial

    Dispatch workflows built on geospatial queries — proximity, coverage and assignment as system state over time.

  • Financial integrity

    A double-entry wallet ledger and serialized payout controls as the foundation for payments.

  • Identity & access

    Role-based access, token rotation and fail-closed OTP across web and mobile.

  • Real-time operations

    Live order and operations state for marketplace, logistics and internal teams.

VIA's website: “Fresh groceries, one tap away”, with App Store and Google Play buttons beside the VIA app icon.
From VIA's public website: the app download section.

Production is part of the product.

I own release governance and executed 144 of 174 production deployments under a gated model: approved operators, green CI, exact-commit verification and forward-deploy safeguards.

A platform-wide security hardening programme of roughly ninety pull requests reduced the public surface to the gateway and strengthened access control, token rotation, OTP, SSRF protection and secret scanning. Reliability baselines run on Prometheus, Grafana, Alertmanager and Sentry, with service health monitoring and restore-tested backups across seventeen databases — and postmortem-driven remediation after a platform-wide outage.

  1. Gate 1Approved operator
  2. Gate 2Green CI
  3. Gate 3Exact-commit verification
  4. Gate 4Forward-only deploy

AI as leverage. Authority stays human.

I established an AI-assisted engineering operating model: written engineering standards, spec-driven agent workflows, architecture and security guardrails, automated verification gates, and human-controlled merge and production approval.

Agents work in parallel inside bounded specs. Architecture, verification and release authority remain with people.

  1. SpecWritten standards and a bounded brief.
  2. Bounded parallel workAgents inside architecture and security guardrails.
  3. VerificationAutomated gates: lint, types, tests, contracts.
  4. Human approvalMerge and production sign-off by people.
  5. ReleaseThrough the same gated path as everything else.

Deferred until justified.

Technology decisions follow cost-based build-versus-buy analysis and phased infrastructure and architecture roadmaps. Several technologies were deliberately deferred until scale justified them.

The aim is a platform that stays understandable to the engineers who inherit it — restraint over noise, architected for the third year.

Next projectEtchilia