Azure + AI architecture

Production-grade AI on Azure, architected properly.

Independent consultancy of Hari S. Twenty years of software engineering, senior Azure architecture, and hands-on AI engineering delivery. London, UK.

Get in touch
Client APIM (AI gateway) AKS workloads Azure OpenAI Entra ID Key Vault Monitor / OTel
  • Engineering since 2005
  • Azure architecture: AKS, APIM, Entra ID, private networking
  • IaC: Terraform and Bicep
  • Microsoft x5, CKAD, Terraform certified

The practice

About

HP Soft Solutions Ltd, trading as ConsultWithCloud, is an independent consultancy. We bring 20 years of software engineering experience, senior Azure architecture expertise, and AI engineering delivery on Azure.

We work as an embedded architect within client teams: designing, building, and handing over systems your engineers can run. Decisions are recorded as ADRs; infrastructure is code; nothing is left as slideware.

Credentials Microsoft Certified: Azure Solutions Architect, Azure Security Engineer (AZ-500), DevOps Engineer (AZ-400), Azure Developer (AZ-204). Certified Kubernetes Application Developer (CKAD). HashiCorp Terraform Associate. Docker Certified Associate.

What we deliver

Services

01

Azure platform and landing zones

Azure platform and landing zones cover AKS, APIM, Entra ID, network topology, and governance. The outcome is a secure foundation your teams can deploy to on day one.

02

AI engineering on Azure

AI engineering on Azure covers Azure OpenAI, RAG, structured outputs, agents, and MCP integration. The outcome is LLM features that survive production, not demos.

03

AI gateway and governance

APIM policies provide routing, rate limiting, and cost controls with identity-first security. The outcome is one controlled front door for every model call.

04

Delivery engineering

Delivery engineering covers Terraform, Bicep, CI/CD, and observability with OpenTelemetry and Azure Monitor. The outcome is that changes ship through pipelines with evidence, not tickets.

Approach

How I work

  1. 01

    Discover

    Current estate, constraints, success criteria.

  2. 02

    Architect

    Options, trade-offs, ADRs, target design.

  3. 03

    Build

    Working software and infrastructure as code, delivered in increments.

  4. 04

    Hand over

    Documentation, runbooks, and your team running it without me.

Selected work

Selected experience

  • Legal-tech SaaS (UK): designed and delivered enterprise MCP governance on Azure as Principal DevOps Architect: APIM as the AI gateway with Entra ID authorization, a governed API Center registry, private networking, and KQL observability across the three-layer MCP error model. A sanitised, Terraform-first reference implementation is public at github.com/collaborationwithothers.
  • Regulated LLM platform: built a data-resident AKS platform for an LLM decision engine with no public endpoints: private endpoints and DNS throughout, secret-free managed identity between services, and Azure AI Foundry model routing for overflow inference.
  • Retail e-commerce at scale: re-platformed a product API from on-premises to Azure, serving 12,000 requests per second at 60 ms during Black Friday peak, and moved the team from one release per sprint to daily production releases.

Enquiries

Contact

Currently taking enquiries for Azure and AI architecture engagements.