Portfolio · Projects
Work.
Production infrastructure, DevOps tooling and SRE platforms built for scale, reliability, and zero-downtime deployments.
deploy.sh
> Initializing build environment...
> Fetching dependencies (128/128)
> Compiling optimized assets...
> Build successful! Deploying to edge...
DevOpsLive
Kubernetes Autoscaler Platform
A production-grade auto-scaling platform on Kubernetes using KEDA and custom metrics, handling 10x traffic spikes with zero downtime.
- Designed and deployed a KEDA-based autoscaling system that reacts to custom Prometheus metrics instead of basic CPU thresholds
- Reduced infrastructure costs by 40% by scaling down idle pods during off-peak hours using time-based and event-driven triggers
- Built Terraform modules for reproducible Kubernetes cluster provisioning on AWS EKS across dev, staging, and production environments
- Configured Grafana dashboards with alerting rules to provide real-time visibility into scaling events and resource utilization
- Implemented Helm charts for all platform components enabling repeatable, versioned deployments across teams
- Achieved sub-30-second scale-out response time through optimised polling intervals and pre-warmed node groups
KubernetesKKEDA
Prometheus
Grafana
Helm
Go
Terraform
CI/CDLive
GitOps CI/CD Pipeline Engine
End-to-end GitOps pipeline using ArgoCD, GitHub Actions, and Helm — enabling fully automated, rollback-safe deployments for 15+ microservices.
- Architected a full GitOps pipeline where all environment state is declared in Git, eliminating configuration drift across clusters
- Integrated Vault for dynamic secret injection, ensuring no credentials are stored in the repository or container images
- Implemented automated canary deployments with Argo Rollouts, enabling progressive traffic shifting with automatic rollback on error budget breaches
- Cut average deployment time from 25 minutes to under 4 minutes by parallelising build, test, and push stages in GitHub Actions
- Set up multi-environment promotion workflows (dev → staging → prod) with manual approval gates for production releases
- Built a custom Slack notification bot that surfaces deployment status, changelog diffs, and rollback options directly in team channels
ArgoCD
GitHub ActionsHelm
DockerKubernetes
Vault
PythonCloudLive
Cloud Cost Optimisation Dashboard
A real-time AWS cost analytics dashboard that surfaces waste, rightsizes recommendations, and enforces budget policies across 6 accounts.
- Integrated AWS Cost Explorer and CloudWatch APIs to aggregate spend data across 6 linked accounts into a unified dashboard
- Built a tagging enforcement system using AWS Config rules and Lambda functions that auto-tags untagged resources and alerts owners
- Implemented rightsizing recommendations using utilisation data from CloudWatch, identifying over-provisioned EC2 and RDS instances
- Achieved a 35% reduction in monthly AWS spend within the first two months by acting on identified savings opportunities
- Created budget alert workflows that notify Slack channels when spend forecasts exceed defined thresholds per team
- Deployed the entire dashboard as a serverless Next.js application on AWS using Lambda and API Gateway for zero-idle cost
AWSPython
PostgreSQL
React
Next.jsTerraform
Lambda
SREIn Progress
SRE Observability Stack
A self-hosted observability platform built on the Prometheus, Loki, and Tempo stack with SLO dashboards and automated incident runbooks.
- Deployed the Prometheus + Grafana + Loki + Tempo stack on Kubernetes using Helm charts with persistent storage and HA configuration
- Configured SLO dashboards tracking error rate, latency (p50/p95/p99), and availability with automatic error-budget burn-rate alerts
- Set up distributed tracing with OpenTelemetry instrumentation across 12 microservices, enabling end-to-end request tracing
- Built centralised log aggregation with Loki and structured log parsing using Promtail DaemonSets on every cluster node
- Created Grafana alert rules with embedded runbook links that guide on-call engineers through incident diagnosis and resolution steps
- Automated platform provisioning using Ansible playbooks, reducing setup time for new environments from days to under an hour
PrometheusGrafanaLLokiTTempoKubernetesHelm
AnsibleKubernetesLive
Service Mesh & mTLS Platform
Zero-trust service mesh using Istio with mutual TLS, circuit breaking, and fine-grained traffic policies across 20+ microservices.
- Deployed Istio control plane in HA mode across 3 availability zones with automatic sidecar injection for all workloads
- Enforced mutual TLS (mTLS) across all east-west traffic with PeerAuthentication and AuthorizationPolicy resources
- Implemented circuit breakers and outlier detection to gracefully handle downstream service failures
- Used VirtualService and DestinationRule resources to orchestrate blue/green and canary traffic shifts with zero downtime
- Integrated Kiali for real-time service topology visualisation and per-route traffic metrics
- Reduced security audit scope by eliminating all unencrypted internal traffic and providing cryptographic service identity proofs
IstioKubernetesEEnvoyKKialiPrometheusJJaegerHelm
CloudLive
Terraform Infrastructure Factory
A modular Terraform monorepo that provisions complete AWS environments — VPC, EKS, RDS, and monitoring — in under 15 minutes.
- Built a layered Terraform monorepo (networking → compute → data → observability) with clear dependency boundaries between layers
- Implemented remote state management with S3 + DynamoDB locking and workspace-based environment isolation
- Integrated Atlantis for pull-request-driven Terraform plans and applies with role-based approval workflows
- Created a module registry with semantic versioning, enabling teams to pin to stable module versions without breaking changes
- Wrote automated Terraform module tests using Terratest, validating real AWS resource creation in isolated accounts
- Reduced environment provisioning time from 2-3 days of manual work to under 15 minutes via the CI pipeline
TerraformAWSEEKSRRDSVVPCGitHub ActionsAAtlantis