Ifeanyi David Ndueze

Cloud & Platform Engineer

Ifeanyi David Ndueze

Cloud & Platform Engineer

Cloud and platform engineer with two decades of infrastructure experience. I build and run production cloud systems end to end — currently a multi-tenant SaaS platform on Linux cloud infrastructure: a nine-service containerised stack with PostgreSQL/TimescaleDB, Redis and Celery, secrets held in AWS SSM Parameter Store, KMS-backed encryption, nginx and Cloudflare strict TLS. Nothing on that server is changed by hand. Hands-on across AWS — CDK, ECS, Aurora/RDS, S3, IAM, CloudFront, SSM, SES, KMS — with GitHub Actions pipelines, Prometheus and Grafana, and delivery across AWS, Azure and Oracle Cloud.

  • Wellington, New Zealand
  • NZ residence visa, full working rights
  • Open to NZ-remote
  • Twenty-plus years in IT infrastructure

Two decades in infrastructure

Every figure on this page comes from a repository or a migration I ran.

  • 9 services in the stack Trading SaaS platform
  • 210,107 rows of market data backfilled Trading SaaS platform
  • 1 TB block volume Cloud hosting stack
  • 300 GB mail migrated Cloud hosting stack
  • SSM Parameter Store
  • AWS KMS
  • Oracle Cloud (Ampere ARM64)
  • Azure Machine Learning
  • Azure AI Foundry
  • HestiaCP
  • GitHub Actions
  • Docker Compose
  • Django migrations
  • pytest
  • restic
  • Backblaze B2
  • imapsync
  • Linux (Ubuntu, ARM64 and x86)
  • nginx
  • Passenger / WSGI
  • PostgreSQL
  • TimescaleDB
  • MySQL
  • MariaDB
  • Redis
  • ADLS Gen2
  • Microsoft Fabric
  • FastAPI
  • Django
  • Celery
  • Next.js
  • Cloudflare (DNS, strict TLS, Email Routing)
  • SPF / DKIM / DMARC
  • RBAC
  • Python
  • TypeScript
  • Bash
  • SQL

What I built, and why

Projects

Trading SaaS platform

Live in paper mode
  • 9 services in the stack
  • 32,926 lines · execution engine
  • 115 commits · web platform
  • 210,107 rows of market data backfilled

The constraint

One ARM64 host in a single region has to run the whole platform, and it holds other people’s exchange API keys.

The decision

Nothing on that server is changed by hand. Every change goes local → commit → push → server pull, and the production schema is committed to git as the single source of truth.

The evidence

  • Deployed as a nine-service Docker Compose stack: postgres, redis, backend, worker, events, beat scheduler, flower, frontend, and the bot itself.
  • Secrets live in AWS SSM Parameter Store, and user exchange keys are sealed with AES-256-GCM under AWS KMS.
  • The production DDL — 71,148 lines — is committed to git as the schema’s single source of truth.
  • The same nine-service Compose stack runs locally as staging, so what is tested is the shape of what runs.
  • Python
  • FastAPI
  • Next.js
  • TypeScript
  • Celery
  • PostgreSQL
  • TimescaleDB
  • Redis
  • Docker Compose
  • Oracle Cloud (Ampere ARM64)
  • SSM Parameter Store
  • AWS KMS
  • nginx
  • Cloudflare (DNS, strict TLS, Email Routing)
  • GitHub Actions

Cloud hosting stack

for a commercial hosting client Infrastructure — no repository
  • 300 GB mail migrated
  • 1 TB block volume
  • 0 mailboxes lost

The constraint

Outbound port 25 is blocked on the cloud provider, and 300 GB of live mail had to move off managed hosting without an outage.

The decision

Migrate incrementally with repeated imapsync passes so the cutover was a short final delta, and route outbound mail through an authenticated relay on 587 with SPF, DKIM and DMARC left intact.

The evidence

  • HestiaCP on an Oracle Cloud Ampere A1 ARM64 instance in Sydney, with a 1 TB block volume.
  • Cloudflare in front for CDN and TLS.
  • Backups run restic to Backblaze B2, and are verified by restoring from them rather than by checking a green tick.
  • HestiaCP
  • Oracle Cloud (Ampere ARM64)
  • Linux (Ubuntu, ARM64 and x86)
  • imapsync
  • restic
  • Backblaze B2
  • Cloudflare (DNS, strict TLS, Email Routing)
  • SPF / DKIM / DMARC
  • Bash

Learning platform

for a mission and training resources centre
  • 10 Django applications
  • ≈20,800 lines

The constraint

Shared cPanel under Passenger and WSGI: no root, no containers, no systemd, no persistent workers — and a live WordPress site whose users had to come across without anyone being asked to reset a password.

The decision

No LMS, CMS or e-commerce library. The quiz engine, grading, progress tracking and certificate generation are all first-party code, because the libraries that would have done that work could not survive the deployment target.

The evidence

  • Django 5.2 on Python 3.11 with MariaDB, deployed to shared cPanel under Passenger and WSGI — no root, no containers, no systemd, no persistent workers.
  • Django
  • Python
  • MariaDB
  • MySQL
  • Passenger / WSGI
  • Django migrations
  • GitHub Actions
  • pytest

Field-services platform

for a cleaning and field-services company in Australia
  • 150 commits
  • 49,760 lines
  • 3 CI workflows

The constraint

The client might not have payment keys configured on the day it went live, and a booking flow that breaks without them is worse than one that never offers cards.

The decision

The platform degrades deliberately: with no Stripe keys configured, card payment is not offered at all and bookings fall back to bank transfer.

The evidence

  • CI proves the WhiteNoise manifest pipeline can hash every referenced static file before anything is deployed.
  • Django
  • Python
  • Django migrations
  • GitHub Actions
  • PostgreSQL

Azure data platform

Coursework
  • 2 Azure tenants

The constraint

The Fabric workspace sat in a tenant provisioned by the lecturer while the storage account was in my own — so identity and access had to be made to work across two tenants.

The decision

Configure the cross-tenant RBAC properly rather than collapsing everything into one tenant, which made the access model real work instead of a diagram.

The evidence

  • A Microsoft Fabric medallion architecture — Bronze, Silver and Gold Delta tables — over an ADLS Gen2 landing zone with RBAC, an orchestration pipeline, an event-driven trigger and Power BI reporting, built on New York high-volume for-hire vehicle trip data.
  • Built across two tenants: the Fabric workspace in a lecturer-provisioned tenant, the ADLS Gen2 storage in my own Azure account, so the cross-tenant identity and access configuration was real work rather than a diagram.
  • Microsoft Fabric
  • ADLS Gen2
  • Azure Machine Learning
  • Azure AI Foundry
  • RBAC
  • SQL

2001 — present

Experience

  • 2022–present

    Independent Cloud & Systems Engineering Consultant

    Wellington, New Zealand

    • Design, build and operate cloud platforms end to end: infrastructure, application, delivery pipeline and deployment are all mine.
    • Replaced a client’s managed hosting with a self-run stack on ARM64 cloud infrastructure, including a 300 GB mail migration completed with no mailboxes lost.
    • Deliver under real platform constraints — shared hosting with no root or containers, blocked outbound SMTP, no persistent workers — and choose the architecture that survives them.
    • Own the CI that gates every deployment, so no change reaches a production server except by commit, push and pull.

  • 2022–present

    Able Seafarer (Deck)

    Coastal passenger ferry operations, New Zealand

    • Concurrent with consulting work, in a safety-critical regulated environment operating 24/7.
    • Procedure is followed without deviation, and operational and defect records are kept accurately to regulatory standard — the same discipline good change management asks for.

  • 2019–2022

    Systems Support Engineer

    Port Harcourt, Nigeria

    Fault diagnosis and resolution across hardware, software and network, with multi-site Windows administration, patching and backup ownership.

  • 2005–2018

    Systems Support Specialist

    Port Harcourt, Nigeria

    A thirteen-year engagement supporting a diverse client base, as trusted point of contact and technical advisor.

  • 2003–2005

    Systems Support Technician

    Abuja, Nigeria

    Led a team of seven technicians across concurrent client projects, running Windows Server and Active Directory environments.

  • 2001–2002

    Systems Technician

    Lagos, Nigeria

    LAN support and application training.

Each one links to where it was used

Capabilities

Cloud Platforms

IaC & CI/CD

Observability & Operations

Networking & Security

Data & Storage

Languages

Application & Web

Formal training

Education & certifications

  • Level 7 Diploma in Cloud Engineering 2026
  • AWS Solutions Architect – Associate (SAA-C03)
  • Simplilearn training certificates AWS Solutions Architect – Associate track, completed 2025
  • CCNA 2010
  • Diploma in Computer Engineering 2000

The same record, a different emphasis

Other versions of this profile

By email or phone

Contact

Open to cloud and platform engineering roles.

david@ndueze.com +64 22 423 3173 github.com/DavidIfeanyichukwu