Open to AI Platform & Cloud Architecture leadership roles

Hi, I'm Harsh.
I design & build

The platforms enterprises run their AI on — landing zones, governance, DevSecOps and the agentic layer on top, delivered end to end for government, BFSI, healthcare and education across the Gulf and India. The architecture and the code that ships it are usually both mine.

Explore my work ↓ ✉ Email me
6+ yrsCloud, DevSecOps & AI delivery
4 cloudsAzure · AWS · GCP · OCI, production
50+ engineersLed on a single program
8 certsMicrosoft · AWS · Google · Anthropic
Harsh Goyal
HG

Harsh Goyal

abu dhabi, uae
Harsh GoyalAI Platform & Cloud Architect
Azure OpenAI
Terraform · GitOps
Agentic · MCP
The stack I ship with, every week
Microsoft Azure
AWS
Google Cloud
Terraform
Kubernetes (AKS · EKS)
Docker
GitHub Actions
Azure DevOps
Claude · MCP
Azure OpenAI
Python
PowerShell
Ansible
Databricks
PostgreSQL
👋 About me

Capability outruns trust. I build the part that closes the gap.

I'm a hands-on cloud and AI platform architect who also leads delivery. Most of my work sits where autonomous systems meet production estates: multi-agent orchestration, tool boundaries, prompt guardrails, approval-gated remediation, and the audit trails that make any of it acceptable to a regulator.

I've designed and built cloud foundations for MOHESR, HCT, Alef Education and Agthia, migrated 500+ Kubernetes workloads to Azure, and shipped three platforms of my own: CloudOps360, an enterprise AI governance framework, and multi-cloud AI-SRE agents for Princess Cruise Lines.

"Equally comfortable in a ministry steering committee and in the Terraform that runs their landing zone."

🎒 What I bring to an engagement

  • ✓
    Foundations built right, onceLanding zones on Azure, AWS and GCP aligned to the Cloud Adoption Framework — identity, network, policy and platform automation as one system.
  • ✓
    Agents that can be trusted in productionTool boundaries, constrained permissions, human approval gates and full audit trails — autonomy you can defend to a regulator.
  • ✓
    I write the code, not just the decksFull-stack contributions across API, identity and frontend on the platforms I architect. Terraform, Python, PowerShell, GitHub Actions.
  • ✓
    Presales to go-live, same person50+ bids won partnering with CEO and sales leadership; then leading 50+ engineers to deliver what was sold.
  • ✓
    Measurable outcomes500+ K8s workloads migrated at ~25–30% lower cost; deployments 3–5× faster; 120+ monoliths modernised for a national airline.
500+Kubernetes workloads migrated to Azure
~25–30%Infrastructure cost reduction on that estate
3–5×Faster deployments via DevSecOps automation
50+Project bids won in a single year
⚙ How I work

Landing zone to autonomous operations

Every engagement runs through the same loop. Click a stage to see what happens inside it.

💼 Presales & solutioning

I win the work before I build it

Solution design and the commercial side of architecture are the first half of every delivery, not a separate team.

🧭

Target architectures

Landing zones, migration roadmaps and AI platform designs sized to the client's real estate, regulator and budget.

📄

RFP / RFI responses

Authored proposals and technical demos tying the solution to executive outcomes — written by the person who'll build it.

🧾

Effort & estimation

Implementation plans, FTE-based effort models and bills of materials, not hand-wavy estimates.

🎬

Working PoCs & demos

CloudOps360 and agentic PoCs demonstrated live against the prospect's own problem. This is usually what wins.

Sold & delivered into:🏛 Government🎓 Education🏦 BFSI🏥 Healthcare🚢 Travel & Hospitality✈️ Aviation
🧠 Capabilities

Six areas I own end to end

From the first landing zone through to agents running inside your estate.

🤖

Agentic AI platforms

Multi-agent orchestration, tool boundaries, guardrails, approval gates and audit trails. Autonomous SRE and cloud operations that regulators accept.

MCPAzure OpenAIClaudeapproval gates
🛡

AI governance

Multi-tenant governance frameworks: AI-scored assessment, server-side verification of model output, versioned contracts, immutable audit.

RBACaudit logversioned scoringconfig-not-code
☁️

Cloud foundations

Enterprise landing zones on Azure, AWS and GCP. Cloud Adoption Framework, hybrid and multi-tenant, migration at national scale.

CAFLanding ZonesAzure · AWS · GCPOCI
⚙️

DevSecOps & platform engineering

Terraform, GitOps, CI/CD, Kubernetes at scale, policy-as-code and secure-by-default patterns.

TerraformGitHub ActionsAKS · EKSpolicy-as-code
🔐

Cloud security

Zero Trust, Entra ID, Sentinel, Defender, WAF and SOC integration. Security designed in from the landing zone, not bolted on.

Zero TrustEntra IDSentinelDefender
🔌

Integration & API platforms

Azure API Management across dev/test/prod — policies, products, lifecycle — including APIM as an AI gateway in front of Azure OpenAI.

APIMOpenAPIBicepAI gateway
🛠 Technical skills

The full toolbox

Everything below has shipped in production; nothing listed for decoration.

Azure

Landing Zones (CAF)AKSAPI ManagementAzure OpenAIEntra ID · PIMAzure PolicySentinel · DefenderKey VaultFunctionsAzure Monitor

AWS & Google Cloud

Multi-account landing zonesEKSIAMCloudWatchCloudFormationGCP org & Shared VPCGKEOCI

IaC & DevSecOps

TerraformOpenTofuBicepGitHub ActionsAzure DevOpsGitOpsAnsiblepolicy-as-codeCI/CD security gates

AI & agentic systems

Multi-agent orchestrationModel Context ProtocolAzure OpenAIClaudeLLMs · RAGGuardrails & evaluationAI governanceAgentic SRE

Engineering

PythonPowerShellAzure CLITypeScriptDockerKubernetesPostgreSQLREST · OpenAPI

Data & security

DatabricksSynapseFabricPower BIZero TrustWAF · SOCFinOpsSAP on cloud
⚡ Engineering highlights

Numbers I stand behind

📦Scale

500+ Kubernetes workloads migrated to Azure on one program

💸Cost

~25–30% infrastructure cost reduction through rightsizing and optimisation

🚀Velocity

Deployments 3–5× faster with Terraform and CI/CD automation

🏗Modernisation

120+ monolithic apps moved to PaaS/SaaS for India's largest airline

🛡Guardrails

Snapshot-protected, approval-gated auto-remediation on every autonomous action

🚀 Selected work

Platforms I've designed & delivered

Production systems inside client estates — under real identity, real governance, real regulators.

SIGNATURE PLATFORM · MULTI-CLOUD · ROLE: ARCHITECTURE, DESIGN & BUILD

CloudOps360 — autonomous cloud operations

The problem. Cloud operations ran on tickets and people: an alert fired, an engineer read it, decided, and typed the fix. Remediation cycles took days and every action was an unlogged judgement call.

The design call. Let AI analyse and propose, never act unsupervised. Every remediation is snapshot-protected and approval-gated, every step is audited, and the same control plane covers inventory, policy and compliance, security, patch management and natural-language-to-Terraform provisioning across Azure and AWS.

  • Control plane: unified estate visibility, compliance state, security posture and patch status
  • Remediation loop: Alert → AI analysis → decision → human approval → remediate → validate → audit
  • Provisioning: natural language to Terraform, reviewed before apply
  • Identity: Entra-authenticated API, scoped admin roles, per-tenant isolation
⚡ Remediation cycles compressed from days to hours · every action explainable
// remediation loop: approval-gated autonomy
Azure estatepolicy · defender · monitor AWS estatemulti-account · cloudwatch Alerts & findingssignals in CloudOps360 control planeinventory · policy & compliance · security · patch AI LAYER · ANALYSE AND PROPOSE, NEVER ACT ALONE AI analysis Proposedaction Humanapproval Snapshot +remediate Validate · audit trail · rollback on failureevery action explainable to an auditor
AzureAWSTerraformPythonEntra IDAzure OpenAIReact
TRAVEL & HOSPITALITY · PRINCESS CRUISE LINES · AWS

Multi-cloud AI-SRE agents

Agentic SRE layer for Princess Cruise Lines' multi-account AWS estate: autonomous signal triage, root-cause analysis and proposed remediation with human-in-the-loop approval — shaped to the client's existing e-mail-driven operating model rather than forcing a new one. Constrained agent permissions, snapshot protection, full audit trail.

🤖 Signal → triage → RCA → approval → remediate → audit
AWSMulti-agentApproval gatesPython
ENTERPRISE · XEBIA · AI GOVERNANCE

Enterprise AI governance framework

Multi-tenant platform for governed AI adoption across service lines: AI-scored maturity assessment with no human scoring, every score re-verified server-side, versioned scoring contracts and pinned model versions, least-privilege RBAC and an immutable audit trail. New business units onboard by configuration, not code.

🛡 Config-not-code onboarding · immutable audit on every score
ClaudeNestJSPostgreSQLEntra IDAKS
EDUCATION · ALEF EDUCATION · AZURE

Landing zone + AWS→Azure migration

On-site design of an enterprise-scale Azure landing zone and lead on the AWS-to-Azure migration for Alef's EdTech platform — Terraform and OpenTofu with GitHub Actions, AKS, high-availability multi-region topology, AI-ready by design.

📦 500+ K8s workloads · ~25–30% lower cost · 3–5× faster deploys
AzureAKSTerraformGitHub Actions
GOVERNMENT · MOHESR · HCT · AGTHIA · AZURE

Cloud transformation & API platforms

Architecture, migration and governance programs for UAE government and enterprise — landing zones, Zero Trust security baselines, and Azure API Management across dev/test/prod for MOHESR and HCT: policies, products, subscriptions, versions and revisions, deployed via Bicep and Terraform.

🏛 Regulated estates · APIM lifecycle as code
AzureAPIMBicepZero Trust
AI GATEWAY · CITADEL ARC · AZURE OPENAI

APIM as an AI gateway

Azure API Management in front of Azure OpenAI and model endpoints — authentication, token-aware throttling, request/response transformation, caching and usage analytics — so application teams consume AI through governed, observable APIs instead of raw keys.

🔌 Governed AI consumption · usage visible per product
APIMAzure OpenAIOAuth2/JWTApp Insights
NATIONAL SCALE · ONDC · PROTEAN EGOV · MULTI-CLOUD

ONDC network & multi-cloud landing zone

Organization-wide GCP, Azure and AWS landing zone and design lead for ONDC — one of India's largest digital commerce networks. 40+ nationally used applications built, secured and operated on Kubernetes with automated CI/CD, Prometheus/Grafana observability and MOSIP identity integration.

🇮🇳 40+ national applications · 15+ engineer team
GCPKubernetesPrometheusMOSIP
OPEN SOURCE · MIT · github.com/harshgoyal1/audit-truth

audit-truth

Can your agent's logs prove what it did? A review skill mapping AI-agent and MCP logging to NIST SP 800-53 AU, ISO 27001, SOC 2 and the OWASP Agentic Top 10 — six auditor questions, code-level recognisers for every gap, and a suggested audit.* record namespace.

📋 Verified against primary sources in two adversarial review rounds
NIST AUISO 27001SOC 2OTel GenAIMCP
OPEN SOURCE · MIT · github.com/harshgoyal1/cloud-architect-skills

cloud-architect-skills

Five reviews a cloud architect actually runs, as AI skills: landing zone (CAF), IaC security, Well-Architected, STRIDE threat model with agentic threats, and FinOps. Evidence-or-nothing, ranked by blast radius, no padding.

🧰 Works with Claude Code, Cursor, Codex, Copilot
CAFWAFSTRIDEFinOps
OPEN SOURCE · MIT · github.com/harshgoyal1/ground-truth

ground-truth

An evidence-based self-review. Points any AI assistant at your own git history, files and sessions, and asks whether what you claim matches what you actually did. Deliberately anti-flattery, local-only.

🪞 Your git history doesn't lie
gitself-reviewprompt
OPEN SOURCE · MIT · github.com/harshgoyal1/cloud-arch-deck

cloud-arch-deck

Architecture diagrams and capability decks as editable PowerPoint, from code. Azure, AWS and open-source stacks with real vendor icons, network boundaries and numbered request paths. Works standalone or as a Claude Code skill.

🖼 ~666 Azure · ~809 AWS · ~3,400 OSS icons behind one command
python-pptxAzureAWSClaude Code skill
🧭 Experience

Where I've built this

Cloud Lead Consultant
Xebia · Dubai, UAE
2025 – PRESENT
  • Lead 50+ engineers across cloud, DevOps, infrastructure, security and AI/ML, owning end-to-end architecture and delivery of a $2M+ multi-cloud transformation program across MEA.
  • Architect and govern Azure, AWS and GCP environments — landing zones, networking, identity, governance and AI/GenAI integration — for MOHESR, HCT, Alef Education and Agthia.
  • Migrated 500+ Kubernetes workloads to Azure at ~25–30% lower infrastructure cost; 3–5× faster deployments through Terraform and CI/CD automation.
  • Designed and built CloudOps360, the enterprise AI governance framework, and multi-cloud AI-SRE agents for Princess Cruise Lines; set up Azure API Management for MOHESR, HCT and Citadel ARC.
Senior Cloud Architect
E7 Group · Abu Dhabi, UAE
2024 – 2025
  • Designed resilient, scalable landing zones with the Cloud Adoption Framework for organization-wide modernization.
  • Led SAP-on-cloud transformation and legacy application modernization; built secure, compliant environments (IAM, encryption, governance) with DR/BCP strategies.
Solutions Architect
Protean eGov Technologies · Pune, India
2023 – 2024
  • Built an organization-wide multi-cloud (GCP, Azure, AWS) landing zone and led the design of ONDC — one of India's largest networks.
  • Developed, secured and operated 40+ nationally used applications on Kubernetes; led incident response and RCA; managed a 15+ member team; integrated MOSIP identity.
Cloud Consultant
Celebal Technologies · Jaipur, India
2020 – 2023
  • Led migration and modernization of India's largest aviation company — 120+ monolithic applications to PaaS/SaaS.
  • Built Azure solutions with security and compliance controls; migrated on-prem Dynamics 365 to online for one of the USA's largest hospitals.
🏆 Recognition & community

Awards, credentials and open source

🏆ORGANIZATION-WIDE AWARD
BEST INFRASTRUCTURE ARCHITECT

Recognised across the organization for infrastructure architecture

Awarded for the design and delivery quality of the cloud foundations behind the programs above.

🏆 Organization-wide
🧠ANTHROPIC · 2026
CLAUDE CERTIFIED ARCHITECT

Certified on designing agentic systems with Claude

Alongside Azure AI Apps & Agents Developer (AI-103) and Cybersecurity Architect Expert (SC-100) — the AI and security halves of the same job.

🎓 3 AI & security certs in 2026
⭐OPEN SOURCE · GITHUB
FOUR MIT-LICENSED TOOLS

Evidence-first review tools for AI assistants

ground-truth (yourself), cloud-architect-skills (your estate), audit-truth (your agents), cloud-arch-deck (your slides) — the same evidence-first stance on four surfaces.

🔗 github.com/harshgoyal1
🎓 Credentials

Certified across the enterprise cloud & AI stack

Claude Certified ArchitectAnthropic · 2026
Cybersecurity Architect ExpertMicrosoft · SC-100 · 2026
Azure Solutions Architect ExpertMicrosoft · AZ-305 · 2026
Azure AI Apps & Agents Developer AssociateMicrosoft · AI-103 · 2026
Professional Cloud ArchitectGoogle Cloud · 2026
Certified Solutions ArchitectAWS · 2025
Azure Security Engineer AssociateMicrosoft · AZ-500
Azure FundamentalsMicrosoft · AZ-900
🎓
B.Tech, Computer Science

Rajasthan Technical University, Kota · 2017 – 2021

📬 Get in touch

Let's build the platform your AI runs on

Agentic AI platforms, cloud security architecture, AI governance — happy to talk through the architecture first. Email is the best way to reach me.

linkedin.com/in/harshgoyal329558· github.com/harshgoyal1· 📍 Abu Dhabi, UAE
Email copied to clipboard ✓