About Sahil Dubey.
Independent DevOps, cloud and compliance architect — I build production AWS and Kubernetes platforms and defend them in the SOC 2, PCI DSS and ISO 27001 room. One accountable owner, 16+ years.
The engineer who builds it and defends it
I'm Sahil Dubey — an independent DevOps, cloud and compliance architect with 16+ years shipping production AWS and Kubernetes platforms, then hardening them to pass the audit. I hold CISA (ISACA) and am an AWS Certified Solutions Architect – Professional. I'm MD/CTO of Techtweek Infotech, based in Mohali, India, and I work remotely with teams across the US, UK, Australia, Singapore, the Gulf and India.
Most teams hire a cloud contractor and a separate compliance consultant, then own the gap between them. I close that gap: one accountable owner who architects the infrastructure and produces the evidence that passes SOC 2, PCI DSS, ISO 27001 and HIPAA.
Selected experience
Cloud & platform
- Migrated 150+ servers from an Auckland colo to AWS Sydney with under 4 hours downtime
- Built and led DevSecOps teams of 12–14 engineers, reporting to CTO/CISO level
- Led the NOC at an AI-lending platform: 99.99% uptime, MTTR down 30%, RTO down 50%
Compliance & audit
- Zero-finding outcomes across SOC 2 Type 2, PCI DSS and ISO 27001 audits
- Risk assessment, evidence automation and auditor liaison end to end
- Regulated experience with fintech, AI-lending and NZ banking clients
Scale
- 200+ projects delivered, 5.0 rating, Top Rated Plus on Upwork
- Managed 6TB+ production MS SQL with tested DR for banking clients
- Modular Terraform across 11+ repositories with GitOps reconciliation
Credentials
- CISA — ISACA
- AWS Solutions Architect – Professional & Associate
- Verify all badges on Credly
Find me on LinkedIn, GitHub, Upwork and Credly, and writing at dev.to, Hashnode and Medium. See how I work and my engagement model.
I run three businesses: Techtweek Infotech for DevOps and cloud engineering, PraxisQ Consulting for compliance and cybersecurity audit, and techtweek.us, which builds custom ordering, booking and portal systems for small businesses across the United States.
Where I'm usually brought in
Platform build and migration
Greenfield AWS landing zones, colo-to-cloud migrations, and the unglamorous rebuild of an environment that grew by hand. Modular Terraform or OpenTofu, one module per concern, state separated per environment, and a pipeline that can roll the change back. EKS where orchestration earns its keep, plain ECS or Lambda where it does not.
Audit readiness, start to signature
Gap assessment against the framework you are actually measured against — SOC 2, ISO 27001, PCI DSS or HIPAA — then the control work, the evidence pipeline and the auditor conversations. The deliverable is not a binder. It is a control that runs on its own and produces its own evidence, so the second audit costs a fraction of the first.
Reliability and cost
Incident response that ends in a fix rather than a dashboard: SLOs that mean something, OpenTelemetry instrumentation, and alerting tied to user impact instead of CPU graphs. On the cost side, Graviton and Spot where they are safe, S3 lifecycle policy, and right-sizing backed by real utilisation data.
Fractional platform leadership
For teams that need senior judgement more than another pair of hands: architecture review, hiring and mentoring, tooling decisions, and the paved road that stops every team inventing its own deployment. I have led DevSecOps teams of 12–14 engineers reporting to CTO and CISO level.
What I work in day to day
Tool choices should be boring and reversible. These are the ones I reach for — and I am equally willing to tell you that you do not need them.
Cloud and infrastructure
AWS across compute, networking, data and identity. Azure where the organisation is Microsoft-first: Azure DevOps pipelines and Repos, AKS and Container Apps, VNet, Entra ID and Key Vault. Terraform, OpenTofu and Terragrunt for the infrastructure itself, Crossplane where a platform team genuinely needs an abstraction over it.
Kubernetes and delivery
EKS and AKS with Helm, and ArgoCD for GitOps reconciliation so the cluster matches the repository rather than someone's laptop. GitHub Actions for CI, with policy as code — Kyverno or OPA — deciding what is allowed to reach production.
Applications and data
.NET, MEAN and MERN, Spring Boot and Java on Tomcat, plus containerising applications that were never built for it. PostgreSQL and Redis by preference; 6TB+ production MS SQL with tested disaster recovery when the client's world is Microsoft.
Observability
OpenTelemetry as the instrumentation standard, Prometheus and Grafana for metrics and dashboards, and eBPF when a question cannot be answered from the application's own logs. Telemetry is written and reviewed like any other code.
Certifications and education
Every badge below is verifiable at source. I would rather you check than take my word for it.
- CISA — Certified Information Systems Auditor, ISACA
- AWS Certified Solutions Architect – Professional, plus the Associate
- Microsoft Certified: Azure Administrator Associate
- B.Tech, Computer Science — Punjab Technical University, 2015
- Diploma in computer hardware and networking — Jetking, 2011
Common questions
Do you work as an individual or through a company?
Both, depending on what the engagement needs. Advisory and hands-on architecture work is usually direct with me. Delivery that needs a team, a contract vehicle or a signed report runs through Techtweek Infotech or PraxisQ Consulting LLP.
Can you issue the certificate or sign the audit report?
No — and be wary of anyone who says they can do both. The certification body or the CPA firm that signs the report has to be independent of the people who built the controls. What I do is get you ready, run the internal audit and manage the external auditor. PraxisQ's SOC reports are signed by a partner US CPA firm for exactly this reason.
Which regions and time zones do you work in?
I am based in Mohali, India, and work with teams in the US, UK, Australia, Singapore and the Gulf. In practice that means a committed daily overlap window with your team, not a promise of 24/7 availability.
What does a first conversation look like?
Thirty minutes, no deck. Your stack, the framework you are being measured against, the date that matters and what breaks if it slips. If I am not the right person for it, I will say so on that call.
Work with one accountable owner.
Book a 30-minute call — tell me your stack, your frameworks and what's on the line.
Book a call