Short, focused
Advisory
A short engagement to review your architecture, security and compliance posture, cloud costs or production operations, and tell you plainly what I find.
Joe Heung Consulting · Dublin, Ireland
Build production systems that stay boring.
Cloud, platform, security and AIOps consulting for technology companies running complex production systems.
I help SaaS and AI companies build, fix and operate platforms that need to be secure, reliable, scalable and cost-effective, and ready for the security reviews and audits that come with enterprise customers. I work on the strategy with your leadership team and on the implementation with your engineers.
Cloud. Platform. Security. Reliability. AI Operations.
- ~20 yearsin technology
- 50+AWS accounts designed & operated
- Multi-regionproduction platforms
- ISO 27001 · SOC 2plus PCI & HIPAA environments
- CTO-levelleadership experience
The pattern
The infrastructure that worked at 10 engineers becomes painful at 50.
Nothing broke all at once. Accounts were added as they were needed. Pipelines were copied from the last project. Alerts were created after each incident and never removed.
Eventually production mostly works, but nobody can explain all of it. Every change feels slightly riskier than it should.
Your AWS environment shouldn't require archaeology every time something breaks.
warnAWS has become complicated. Nobody is sure which accounts, roles and VPCs still matter.
warnNobody fully understands production. The knowledge lives in two people's heads.
warnAlerts are noisy, and the on-call rota has learned to ignore them.
warnCloud bills keep climbing, and nobody can say which workloads are responsible.
warnSecurity questionnaires expose gaps, and enterprise deals wait on the answers.
warnAn auditor wants evidence of access reviews, and nobody knows where it lives.
warnReleases feel risky, so they happen less often and get bigger.
warnOwnership between engineering and operations is unclear.
warnAI workloads arrive with problems nobody has solved here yet: GPUs, model versions, evaluation, cost.
infoI help put in place the architecture, processes and visibility you need to get back in control, then hand it back to your team.
Services
Five areas. Usually more connected than they look.
Reliability problems often turn out to be architecture problems. Security gaps often turn out to be ownership gaps. I work across all five so the fixes line up.
01
Platform & Cloud Architecture
Build a platform your engineering team can actually operate.
Design or modernise production infrastructure on AWS, from account structure, networking and containers to data and delivery, so it scales without turning into a maze.
Details02
Cloud Security & Compliance
Security that works with engineering rather than against it.
Cloud security engineering plus the technical side of ISO 27001, SOC 2, PCI and HIPAA, so you can pass enterprise security reviews and audits without stalling the roadmap.
Details03
Production Reliability & Operations
Know what is happening in production before your customers tell you.
Observability, alerting, incident process and operational ownership, so production stops surprising you.
Details04
AI Infrastructure & AIOps
Turn AI experiments into production systems you can operate confidently.
GPU infrastructure, model serving, evaluation and observability for AI workloads, designed around real usage patterns and real costs.
Details05
Fractional Technology Leadership
Senior technical leadership without needing another full-time executive.
A fractional CTO or Head of Platform, Cloud or Security. Experienced leadership while you build permanent capability.
Detailsnot sure where it fits?
Problems rarely arrive neatly labelled
An outage turns out to be an ownership problem. A security gap turns out to be an architecture problem. Describe what's going on and I'll tell you honestly where I can help.
Get a second opinionSecurity & compliance
Pass the security review without stalling the roadmap.
Selling to enterprise, financial-services or regulated customers brings questionnaires, audits and due diligence. I help build the controls into the platform itself: access, encryption, logging, recovery and guardrails in code. The answers are then true, and the evidence comes straight from the platform.
- iso 27001Experience working on ISO 27001 programmes, from scope and controls to evidence.
- soc 2Technical readiness, with controls that generate their own evidence.
- pci · hipaaScoping, segmentation and safeguards for payment and health data.
- assuranceSecurity questionnaires, customer reviews and vendor due diligence.
- resilienceRansomware readiness, tested backups and disaster recovery.
The difference
Strategy without losing sight of the implementation.
I can help a leadership team decide how production operations should work. Then I can design the AWS architecture, write the Terraform, build the pipelines and set up the observability and security controls to make it real. With one person doing both, nothing gets lost between the plan and the build.
In the leadership meeting
- Production operating model and ownership
- Cloud and platform roadmap
- Security posture, risk and compliance readiness
- Answering enterprise customers and auditors
- Build, buy or hire decisions
- Vendor selection and cloud cost control
- Technical due diligence
In the terminal
# same engagement, same person
$ terraform plan -out=network.plan
$ gh workflow run deploy.yml -f env=staging
$ aws iam simulate-principal-policy …
$ openssl x509 -noout -enddate -in api.pem
$ otelcol validate --config=collector.yaml
$ aws backup start-restore-job …
$ vllm serve ./model --max-model-len 8192How I work
From architecture diagram to production.
-
stage 01
Understand
Start with how your product and engineering organisation actually work: the systems, the people, the incidents and the constraints. Not how the wiki says they work.
-
stage 02
Simplify
Remove infrastructure and process complexity that isn't earning its keep. Fewer moving parts are easier to secure, cheaper to run and faster to debug.
-
stage 03
Build
Make practical improvements in code, pipelines and configuration. I don't produce architecture decks that nobody executes.
-
stage 04
Transfer
Leave your engineering team able to own the system: documented, understood and operable without me.
Representative experience
Around two decades on the systems underneath the product.
Startups, scale-ups and enterprise, from joining an early-stage SaaS company as one of its first DevOps engineers to CTO-level responsibility.
aws / multi-region
Multi-account SaaS infrastructure
Designed and operated infrastructure spanning more than 50 AWS accounts across multiple regions, including account strategy, networking, automation and the operating model around it.
ha / real-time
High-availability platforms
Designed and operated highly available production workloads, including real-time systems where latency and availability are immediately visible to users.
security / governance
Enterprise security
Worked with financial-services and enterprise organisations where security, reliability and auditability are requirements, not preferences. Hands-on experience of ISO 27001 programmes, plus SOC 2, PCI and HIPAA environments.
gpu / inference
AI infrastructure
Designed deployment approaches for GPU-backed inference and fine-tuning workloads using vLLM and QLoRA, including AWS and GCP architecture comparisons and cost analysis.
cto / platform / security
Technology leadership
Roles spanning hands-on engineering, platform leadership, security leadership and CTO-level responsibility, from very small startups to enterprise environments.
Ways of working
Sized to the problem, not to a retainer.
Hands-on delivery
Project
Hands-on delivery of a defined technical outcome. I design it, build it with your team, and leave it documented and owned.
Ongoing leadership
Fractional
Ongoing senior leadership, typically one or more days a week. I join the leadership or engineering organisation while you build permanent capability.
A second pair of hands
Troubleshooting
For difficult infrastructure or production problems where your team would benefit from another experienced engineer working alongside them.
Next step
Have a difficult infrastructure problem?
Tell me what you're building, what isn't working, or where you need another experienced pair of eyes.
Discuss a project
No sales process. You'll hear back from me directly.