Skip to content
UJJAVAL PARMAR
Get in touch

[ infrastructure & delivery ]

I keep deployments boring.

Associate DevOps Engineer at GlobalVox. I build the pipelines, clusters and monitoring that let five microservices ship on a Friday without anyone watching a dashboard.

Read a case studyDownload résumé

Ahmedabad, Gujarat, India · open to remote

PIPELINE · payments-apiPASSING
  • build → docker hub42s
  • helm values bump8s
  • argocd sync · staging1m 04s
  • rolling update · prod3/5 pods
downtime this deploy0.00s
50%
Cut in deployment time
99.99%
Uptime SLA on multi-AZ AWS
40%
Faster incident debugging
0
Downtime incidents in 90 days

[ selected work ]

Four systems, and what they cost to run.

Each one opens into a full writeup: the constraint, the architecture, what broke first, and the numbers after.

0150% faster deployments

Automated GitOps CI/CD with Kubernetes & ArgoCD

A multi-branch Jenkins pipeline that updates manifests and image tags itself, so five microservices deploy without anyone editing YAML.

  • Jenkins
  • ArgoCD
  • Kubernetes
  • Helm
  • Doppler
Read the writeup →
0299.99% uptime SLA

Scalable & secure AWS architecture with load balancing

Multi-AZ VPC, application load balancing and scoped security groups, provisioned end to end in Terraform so the whole environment is reproducible.

  • AWS
  • Terraform
  • VPC
  • ALB
  • Route 53
Read the writeup →
0340% faster debugging

Centralized monitoring & logging infrastructure

ELK for logs, Prometheus and Grafana for metrics, New Relic for traces. One place to look when something is slow at 2am.

  • ELK Stack
  • Prometheus
  • Grafana
  • New Relic
Read the writeup →
04Zero downtime

Kubernetes high availability cluster

Multi-node clusters with rolling updates, liveness and readiness probes, pod disruption budgets and automated rollback on a failed check.

  • Kubernetes
  • EKS
  • Helm
  • Docker
Read the writeup →

[ how it works ]

From commit to production, with no human in the path.

  1. STEP 01

    Build

    Jenkins builds the image, tags it by commit and pushes to Docker Hub with version retention.

  2. STEP 02

    Declare

    The pipeline updates Helm values in the manifest repo. That commit is the deploy request.

  3. STEP 03

    Reconcile

    ArgoCD detects drift and syncs the cluster, scoped by RBAC per environment.

  4. STEP 04

    Verify

    Probes and Grafana alerts decide. A failed check rolls back before anyone pages me.

[ the shift ]

What changed when I owned the pipeline.

Same team, same services. The difference is what a release asks of a person.

BEFORE

  • Manifests edited by hand, image tags drifting between environments.
  • Secrets pasted into CI variables, with rotation nobody tracked.
  • Onboarding a service meant copying last week's YAML and hoping.
  • Debugging started with SSH and ended in three unrelated log files.

AFTER

  • Git is the only source of truth. ArgoCD reconciles, RBAC scoped per environment.
  • Doppler holds every secret. Nothing sensitive lives in a pipeline config.
  • One reusable Helm chart. A new service onboards in an afternoon.
  • ELK, Prometheus and Grafana. The dashboard answers before I do.

[ stack ]

Tools I run in production, not just in tutorials.

Hiring for infrastructure that has to stay up?

I reply within a day. Happy to walk through any of the case studies in detail.

ujjavalworkmail22@gmail.com