I build Agentic DevOps and AI ready adaptive infrastructure, with security built in.

Since 2018, with 8+ years of experience. From self managed Kubernetes to serverless platforms,
I automate operations so your team can focus on building the product. Enterprise infrastructure on AWS, GCP and Azure, bare metal servers from top providers, and on premises hosting, engineered for 99.9% uptime with IDS and IPS from the start.

Hassan Khurram
Hassan KhurramDevOps and Platform Engineer

Professional Cloud Architect

What is DevOps?

DevOps is a cultural and organizational philosophy, not a toolset or a title.

It aligns people, process and technology so software teams deliver reliable value faster.

Google calls its version SRE. The SRE Workbook describes DevOps as a wider philosophy and culture, and SRE as one way to put it into practice: class SRE implements interface DevOps.

Next

The DevOps loop. Development on the left: plan, code, build, test. Operations on the right: release, deploy, operate, monitor. Then back to plan.PlanCodeBuildTestReleaseDeployOperateMonitorDevOps
The loop does not end. What monitoring finds is what gets planned next.

Why DevOps?

DevOps brings development and operations together through shared ownership, automation, clear responsibilities, and continuous feedback, to deliver software faster and more reliably.

Put structure around delivery, and the same engineers ship more often, with less rework and fewer surprises in production.

Measured the way DORA measures delivery: deployment frequency, change lead time, change fail rate, failed deployment recovery time and deployment rework rate.

Next

Core message breakdown

A four panel drawing: siloed teams dragging releases by hand, then everyone doing everything at once, then no automation with developers fixing servers, then a DevOps flow where automated pipelines let each role do its own work
  1. Siloed teams

    The cost of silos

    When leadership or product teams simply dictate direction without shared ownership, Dev, QA and Ops are forced to manually drag heavy releases forward in isolated steps. Work is slow, exhausting and disconnected.

  2. Everyone does everything

    The trap of everyone does everything

    Removing silos without defining clear boundaries creates confusion. When engineers are tangled in laptops, wrenches, cables and blueprints all at once, context switching slows down delivery and tasks stay unfinished.

  3. No automation

    The pain of zero automation

    Without automated pipelines and clear documentation, people end up doing the wrong work. Developers get stuck fixing servers, Ops struggles to decipher unclear customer requirements, and the team physically carries a broken system on its back.

  4. DevOps flow

    True DevOps flow

    When automation handles repetitive infrastructure and testing, every role can focus on what it does best. Sales communicates clear requirements, Dev ships code smoothly, Audit verifies compliance, Ops and SRE optimise the cloud engine, and product and leadership move forward.

When DevOps?

As early as you can. Pipelines, environments, access and ownership are cheap decisions at the start of a system and expensive ones in production.

Left until later they get made one incident at a time, and that is how a platform turns into spaghetti.

Put them in early and the system stays aligned as it grows. The chart is the same move at industry scale.

Google's SRE book reaches the same conclusion in its chapter on the early engagement model: the earlier the consultation, the better the service, because design changes made once it is running carry a high cost in effort and complexity.

Next

A diagram in three zones: manual operations, automated delivery, then agentic operations. Two curves, resources and cost to run and maintain, fall from high to low across them. A third curve, output shipped, rises from low to high and crosses both025507510020102020203020402050

Illustrative. Relative levels and an industry timeline, not measured data.

Platforms and tools

By the numbers

  • 8+Years in productionSince 2018
  • 30+Projects deliveredPublic and under NDA
  • 3Major cloudsAWS, Google Cloud and Azure
  • 5OrganizationsSingapore, the US and Pakistan

Cloud platforms and hosting

  • Google CloudInfrastructure and DevOps
  • awsAWSInfrastructure and DevOps
  • AzureAzureCloud infrastructure
  • CloudflareSecurity, DNS, serverless
  • VercelFrontend hosting
  • HetznerBare metal servers
  • ScalewayBare metal and cloud
  • ContaboBare metal and VPS
  • OneProviderBare metal servers
  • HostingerManaged hosting
  • MongoDB AtlasManaged MongoDB
  • NeonManaged Postgres
  • UpstashServerless Redis

SaaS and tools

  • GitHubCode and CI/CD
  • GitLabPipelines and registries
  • BitbucketClient pipelines
  • Docker HubImage registry
  • TerraformInfrastructure as code
  • AnsibleConfiguration management
  • DockerContainers and signing
  • KubernetesSelf managed clusters
  • ProxmoxBare metal virtualisation
  • WatchtowerScoped container updates
  • TailscalePrivate fleet networking
  • NetBNetBirdMesh VPN on WireGuard
  • WireGuardEncrypted tunnels
  • MikroTikRouting and VLANs
  • NGINXReverse proxy and TLS
  • GrafanaDashboards and alerting
  • PrometheusMetrics and alert rules
  • OpenTelemetryInstrumentation
  • SentryError tracking
  • New RelicApplication monitoring
  • DatadogMetrics and APM
  • SplunkLog search and analysis
  • Fluent BitLog collection
  • BigQueryAudit log retention
  • DiditDiditIdentity verification
  • share2.usEncrypted file sharing
  • MySQLDatabases and upgrades
  • PostgreSQLDatabases and backups
  • MongoDBDocument stores
  • RedisCaching and rate limits
  • RabbitMQQueues and messaging
  • GoServices and CLI tools
  • TypeScriptServices and tooling
  • PythonAutomation and tooling
  • PHPLegacy services
  • JavaJVM services
  • BashScripting and runbooks
  • Claude CodeAgentic DevOps
  • Gemini CLIAgentic workflows
  • CodexCodexAgentic development
  • MCPTools agents can call
  • UbuntuProduction servers
  • DebianServers and containers
  • Rocky LinuxRHEL compatible hosts
  • Red HatEnterprise Linux
  • OpenBSDBSD systems
  • SendGridSendGridTransactional email
  • MailgunEmail delivery
  • TeamsMicrosoft TeamsTeam communication
  • Google WorkspaceMail and collaboration
  • ThunderbirdMail client
  • DocuSignDocuSignElectronic signatures
  • SignDeskSignDeskeSign and KYC workflows
  • DowDow JonesRisk and screening data
  • W-ChkWorld-CheckSanctions and PEP data
  • Google Custom SearchProgrammable search
  • ClaudeAnalysis and drafting
  • ChatGPTChatGPTResearch and drafting
  • OpenRouterOne API, many models
  • Hugging FaceModels and datasets
  • AgoraReal time AI and video
  • Vertex AIAgent Builder on GCP

About me

I build and run production infrastructure: cloud platforms, release pipelines, security and recovery, for teams that need all of it to work every day.

Google Developers Group introduced me to Google Cloud at university, and it stuck. I finished my bachelors in 2020, started in DevOps at HeuristicSol in early 2021, and certified as a Google Cloud Professional Cloud Architect in 2024.

The work is enterprise grade production: databases, web servers and object storage at scale, signed release pipelines, restores during incidents, and live incident response handled by AI agents with a human in the loop whenever it matters.

I mentor the engineering team at HeuristicSol, have lectured on agentic development and agentic DevOps at the University of Management and Technology, and am part of the team behind Google Developers Group events in Lahore.

Speaking and community events, Lahore

Academic qualifications

Google Cloud Certified Professional Cloud Architect badge

Projects

Production systems I designed, built and operated.

2026

SplashArk

Social video and marketplace platform on AWS

2026 to now

share2.us

A2A & E2E easy file sharing. Founder, open source, semi open source
GitHub

2025

ButterPOS

Multi tenant restaurant point of sale platform

2025

Klok HR

Multi tenant HR and payroll platform

2018 to now

OrthoSuite

Multi tenant clinical operations platform

2025

Watchtower

Agentic AI support dashboard, built in house

One Constellation

One Constellation

Financial compliance and onboarding platform 2024 to now

Visit one-constellation.com

Cloud infrastructure and security for a financial compliance platform. KYC, screening, transaction monitoring and shared single sign on across three Google Cloud projects and seven environments.

  • Build and operate the Google Cloud estate behind Node, PHP and Go services across seven environments.
  • Designed the platform security architecture: load balancer fronted ingress, a modern TLS floor, and security headers and CORS in application middleware, so controls move with the service rather than the cloud.
  • Built distributed rate limiting on a shared store, verified under load to hold across every running instance.
  • Produced SOC 2 Type II evidence across three consecutive annual cycles. Audit logs retained in BigQuery, with a reproducible trail across GitHub, Google Cloud and Jira.
  • Rehearsed the Cloud SQL upgrade from MySQL 8.0 to 8.4 on a production sized clone. 13 minutes end to end, 3.5 minutes of downtime.
  • Introduced reference aware lifecycle cleanup for the container registry and Cloud Run revisions. 509 GB reclaimed.
  • Designed selectable static egress IPs on WireGuard and Cloud NAT across two regions, failing closed rather than leaking to a default address.
  • Led the organisation wide Node runtime migration across 155 workflows and 66 repositories, verified against the serving image rather than the declared config.
  • GCP
  • Cloud Run
  • Cloud SQL
  • Terraform
  • GitHub Actions
  • WireGuard
  • BigQuery
SplashArk

SplashArk

Social video and marketplace platform on AWS 2026

Visit splashark.com

Staging and production AWS for a social video platform with live streaming, a marketplace, an advertising platform and creator payouts. Sole operator across about 14 microservices, two web frontends, a mobile app and two AI services, with every production change signed off by the founder.

  • Moved the platform from EC2 to ECS Fargate in two weeks, with zero downtime in production.
  • Own the Terraform for both environments: VPC, load balancer, ECS services, Lambda and the parameters behind every microservice. Moved state to a shared S3 backend with locking, verified by matching plans before and after with zero drift.
  • Made Cloudflare R2 the single media origin across all 12 backend services, with all 12 deployments shipped green.
  • Designed and built a scheduled task platform from scratch on EventBridge Scheduler and ECS, running the five tasks behind the Creator Pay revenue share, with its own minimal IAM role and CloudWatch logging. The same pattern now runs an hourly referral reward check.
  • Payments on Stripe and PawaPay: two correctly scoped webhook endpoints, one for platform events and one for connected accounts, with secrets delivered through SSM and verified with a real transaction.
  • Production releases as isolated cherry picks rather than branch merges, every deploy gated on health checks with automatic rollback. One campaign promoted ten workstreams this way.
  • Least privilege throughout: OIDC deploy roles in GitHub Actions, secrets in SSM Parameter Store, and a read only database role for internal tooling with both the allowed read and the blocked write verified.
  • AWS
  • ECS Fargate
  • Terraform
  • GitHub Actions
  • Cloudflare R2
  • EventBridge
  • Stripe
  • PawaPay
  • MongoDB
  • Redis
  • RabbitMQ
  • Bedrock
share2.us

share2.us

A2A & E2E easy file sharing. Founder, open source, semi open source GitHub 2026 to now

Visit share2.us

End to end encrypted file sharing for the command line, the desktop and AI agents. One share, three surfaces, one account.

  • Designed cross account agent handoffs that hold even against a compromised server: signed hops with a freshness window and nonce replay protection, pinned device keys, and delivery bound to a session the receiver derives independently.
  • Built peer to peer transfer over LAN, Tailscale and WireGuard using PAKE and TLS 1.3. Direct path first, relay only as a last resort, so most transfers never reach a server.
  • One shared core library underpins both the CLI and the desktop app, preventing behavioural drift across surfaces. Native installers for three operating systems.
  • Safe defaults are not overridable by accident. A local secret scan runs before every upload, and high risk options are excluded from standing defaults by design.
  • All clients are GPL-3.0. Code that touches user files runs on the user machine and stays open; the hosted service funds the work.
  • Go
  • TypeScript
  • Wails
  • MCP
  • Cloudflare
  • PAKE
  • TLS 1.3
ButterPOS

ButterPOS

Multi tenant restaurant point of sale platform 2025

Visit butterpos.com

Cloud and on premises infrastructure for a multi tenant point of sale platform. Over 60 tenants, and a containerised stack running offline inside each restaurant.

  • Owned the platform end to end: 60+ tenants across four production stacks, plus a 21 container offline stack at more than 20 outlets.
  • Moved the platform from paid VPS hosting to Google Cloud, where it scales with demand, balancing reliability against cost.
  • Designed and built the fleet control plane: an outlet registry as single source of truth, desired state per server, update orchestration and full deployment history.
  • Built a signed release pipeline. Images signed with Cloud KMS, verified by signature and digest on each server before update, with the control plane treated as untrusted.
  • Built nightly off site backups for 61 tenant databases on a pull model, with retention, weekly verification and dead man switch alerting. Restores proven at 205 tables and 154,393 rows.
  • Built an AI operations agent that diagnoses outlets over a private network. Human approval required for critical changes, cryptographically bound to the exact command permitted.
  • Defined RTO and RPO for business critical services: one hour recovery, near zero data loss on financial data, backed by quorum queues.
  • Standardised container builds and CI/CD across the estate, and defined the logging, metrics and tracing stack used to operate the fleet.
  • Automated tenant domain provisioning and moved per tenant frontends to managed hosting, removing manual onboarding.
  • GCP
  • Docker
  • Proxmox
  • Ansible
  • Tailscale
  • cosign
  • Grafana
  • New Relic
Klok HR

Klok HR

Multi tenant HR and payroll platform 2025

Visit demo.klokhr.com

Infrastructure, delivery and server management for a multi tenant HR and payroll platform. A workspace per tenant on its own domain, provisioned automatically when the tenant is created.

  • Moved both frontends onto managed edge hosting, each tenant on its own domain with its own certificate. The migration tool is a dry run unless told otherwise, backs up every record it touches and can roll a whole run back. Each tenant went live in 50 to 90 seconds.
  • Built tenant domain provisioning into the backend: creating a tenant attaches its domain and writes the DNS record, deleting it releases both. Off until configured, credentials proven before anything is written, and a failed attempt rolls back only what that attempt created. 12 unit tests against an in memory API.
  • Moved the tenant APIs onto a current PHP runtime inside a container, with the host packages left in place to roll back to. Proven before the cutover by replaying 257 authenticated routes against the old runtime and comparing status and body: 257 of 257 identical.
  • Separated staging from production end to end: its own private image repository, its own credentials on the host, and its own scoped updater, so a staging deploy cannot reach the production stack. A push reaches staging inside five minutes.
  • Built the production image pipeline on GitHub Actions, tagging every build by commit so a release can be named, found and rebuilt.
  • Kept the serving edge current across the backend containers, recreating only the service that changed and checking the result against a baseline captured before the change.
  • Run the servers underneath it: provisioning, access, day to day operation, and a backup taken before every change with a path back to it.
  • Ubuntu
  • Docker
  • Ansible
  • GCP
  • Cloudflare Pages
  • GitHub Actions
  • Laravel
  • MySQL
OrthoSuite

OrthoSuite

Multi tenant clinical operations platform 2018 to now

Visit portal.orthosuite.net

Infrastructure on Google Cloud for a multi tenant clinical operations platform, built and still operated eight years on. Serverless web services, managed databases, and backups held outside the provider that runs it.

  • Built the Google Cloud estate it runs on: App Engine and Cloud Run for the web services, Compute Engine for the databases, and Cloud Storage for the files.
  • Delivery runs from GitHub through Cloud Build triggers, so every release reaches the platform the same way.
  • Backups leave the provider. They are written to object storage outside Google Cloud, so a bad day inside one account does not take the copies with it.
  • Operated to SOC 2 practices, with secrets encrypted at rest using SOPS so configuration can live in version control without the keys living there too.
  • GCP
  • App Engine
  • Cloud Run
  • Compute Engine
  • Cloud Build
  • GitHub
  • SOPS
Brighten Learning

Brighten Learning

Social skills and career readiness curriculum 2022

Visit brightenlearning.com

Server operations and delivery for an animated, interactive learning platform. Two VPS servers behind a load balancer, released through A/B deployments from Bitbucket Pipelines.

  • Two Ubuntu VPS servers on Hostinger behind a load balancer, released through A/B deployments, so one side takes the new version while the other carries the traffic.
  • Delivery from Bitbucket Pipelines to the servers, with PM2 running the Node services and bringing them back when they stop.
  • Access over SSH only, from allow listed IP addresses.
  • Monitoring on Grafana across the servers and the services on them.
  • Maintained the estate, deployed new sites onto it through PM2, and debugged the services in production.
  • Ubuntu
  • Node.js
  • C#
  • MySQL
  • Bitbucket Pipelines
  • PM2
  • Hostinger
  • Grafana
Watchtower

Watchtower

Agentic AI support dashboard, built in house 2025

An agentic AI support dashboard for ButterPOS. When a restaurant has a live problem, an AI agent investigates it and proposes the repair, and a person approves it with one look at exactly what it will do. Support fixes the outlet over the private network, without anyone having to be free to open a remote session, so the team can spend its time on what actually needs a person.

  • The agent is built on HolmesGPT with a custom ButterPOS operations MCP server and Gemini. It investigates outlet and server problems with read only tools: logs, containers, health checks, queues. Its conversations are kept in MongoDB.
  • It proposes repairs; it does not apply them. Every fix goes to a superadmin as an approval card that shows exactly what the fix does, and the approval is signed and tied to that action and its arguments, which the repair server verifies independently before anything runs.
  • Live issues open support cases on their own, and the WhatsApp alert links the operator straight to the case, already signed in on the right page. Team messaging and scheduled work are tracked in the same place.
  • AI usage is metered: token and cost tracking for every AI feature, so the agent's running cost is a number on a page rather than a surprise on an invoice.
  • Underneath it, the operations console: every outlet's version, reachability and container state, collected every 5 minutes, and outlet updates that confirm they landed against the running containers and the health check rather than being assumed.
  • Reachable only over the company's Tailscale network, with sign in against Watchtower's own OAuth 2.0 and the session kept in an encrypted, HttpOnly cookie. Role based access with every action audited, and the dashboard's own server can reach the fleet console but never an outlet directly.
  • Designed and built the dashboard, the fleet console and the agent, and run the infrastructure under them: servers, private network policy, certificates, deploys and monitoring.
  • HolmesGPT
  • MCP
  • Gemini
  • Next.js
  • TypeScript
  • Cloud Run
  • Cloud SQL
  • MongoDB
  • Postgres
  • Tailscale
League of Nations

League of Nations

Geopolitical analysis platform 2024

Visit leagueofnations.co.uk

Infrastructure for League of Nations, a geopolitical analysis platform, built for Hogan Solutions. Set up end to end on a VPS.

  • Set up the infrastructure for the site and its API on a VPS.
  • Delivered for Hogan Solutions, led by CEO Brendan Hogan.
  • VPS
  • Next.js

Organizations

Happily worked with the multiple organizations globally.

  1. HeuristicSol Pvt Ltd
    • Senior Platform Engineer
    Feb 2021 to nowFull time

    HeuristicSol Pvt Ltd

    Singapore

    HeuristicSol places engineers with client companies. Through them I work with Ascent Fintech Solutions in Singapore, owning DevOps and cloud security across their Google Cloud estate. Alongside that, CI/CD and infrastructure as code for HeuristicSol itself, and mentoring the engineering team.

  2. Bizmatic Technologies
    • Platform Engineer
    • Cloud Architect
    • Technical Lead
    Apr 2025 to Dec 2025Contract

    Bizmatic Technologies

    United States

    Owned the full lifecycle of cloud and DevOps for their point of sale and financial services platform. Infrastructure, deployment automation, observability and reliability, plus the logging service built to meet their compliance requirements.

  3. SplashArk Inc
    • DevOps Engineer
    • Infrastructure Engineer
    Apr 2026 to nowContract

    SplashArk Inc

    Run staging and production AWS for a social video and marketplace platform, sole operator across about 14 microservices, two web frontends, a mobile app and two AI services. Terraform, ECS Fargate, GitHub Actions with OIDC deploy roles, and payments on Stripe and PawaPay.

  4. ECrunch Tech
    • Technical Lead
    • Backend Engineer
    Feb 2020 to Nov 2022Contract

    ECrunch Tech

    Singapore

    Led a small team integrating new features. Cut API response times substantially with Redis caching, refactored Dockerfiles and replaced vulnerable base images.

  5. Innovative Health Concepts
    • DevOps and Assist. Director Of Information Technology
    Feb 2021 - NowContract

    Innovative Health Concepts

    Lahore, Pakistan

    Built and supported the internal systems and server infrastructure, including a PHP and MySQL records platform that replaced manual data collection, and the on premises hardware and network it ran on.

Services

Platform, security and reliability engineering.

  • Platform and DevOps ownership

    Cloud infrastructure, release pipelines and environment isolation, kept predictable as the system scales. Includes leading engineers through the migration and documenting it for the people who have to act on it.

  • Hardening and security remediation

    Bringing infrastructure already in production to a defensible baseline, and converting external assessment results into shipped fixes. A WAF in front of what is exposed and host intrusion detection on the machines that matter. Rollout sequenced so one environment proves the pattern, then delivered everywhere with a backup and per service rollback.

  • Identity and least privilege

    Machine clients moved to narrowly scoped identities and keyless, impersonation only deploy roles. OAuth 2.0 and JWT at the service boundaries, TLS and PKI managed rather than assumed. Long lived keys audited and retired once proven unused. Secrets moved behind a manager rather than environment variables.

  • Audit and compliance evidence

    SOC 2 and audit readiness. Retained audit logs, reproducible evidence trails, and the automation that makes controls demonstrable on demand rather than annually.

  • Reliability and disaster recovery

    Backups proven by restore, an agreed RTO and RPO, and release images signed and verified on the receiving end, so a server installs only what it can prove came from the expected build.

Client feedback

  • Hassan has run our AWS infrastructure since April 2026, staging and production. He took us from EC2 to ECS in two weeks with zero downtime in production, and moved all our media onto Cloudflare R2 across twelve services without a single failed deploy. He is reliable, careful and quick, and genuinely passionate about the work he does with us.
    Marvin MvondoCEO, SplashArk Inc
  • Hassan has been a balanced architect in terms of reliability and cost. He shifted us from paying for VPS to scaling our application in Google Cloud effortlessly. He ran ButterPOS's cloud and DevOps end to end, across more than 60 tenants and every restaurant's offline stack. Nightly backups proven by real restores, signed releases to every outlet, and a control plane for the whole fleet. Dependable, thorough, and always a step ahead of the next problem.
    Saib KhanCEO, Bizmatic Tech.
  • Hassan played a vital role as our Backend Developer and DevOps Specialist at ECrunchTech. With a proactive approach and strong problem-solving skills, Hassan optimised performance and streamlined our deployment processes. A top choice for any organisation.
    Romaisa RiazCEO, ECrunchTech
  • Hassan set up the infrastructure behind League of Nations in 2024 and got it live quickly and cleanly. He has a quick mind and knows how to untangle spaghetti code. Clear communication, solid work, and a setup that has simply kept running. I would happily work with him again.
    Brendan HoganCEO, Hogan Solutions