// projects.impact — measured
Proof of Work.
Real systems shipped at CIBC (banking) and Tata Play (DTH media). Every number below ran in production — speed up, cost down, reliability up.
Istio Traffic Mirroring & Canary Deployments
Mirrored live traffic for zero-risk prod testing — MTTD ↓40%, MTTR ↓35%, mTLS locked on every service pair.
✗ Before
- Big-bang releases — 100% traffic switches at once
- No production traffic testing before go-live
- Plaintext service-to-service traffic
✓ After
- Mirrored live traffic for zero-risk prod testing
- Canary rollouts — 10% → 50% → 100% gradually
- mTLS locked on every service pair
↑0downtime releases
↓40%detect time
↑100%mTLS traffic
Istio
RHOCP 4.x
mTLS
Helm
HashiCorp Vault Enterprise Secrets Platform
Every secret short-lived and auto-rotated — 0 hardcoded credentials left in CI/CD or Kubernetes.
✗ Before
- Plaintext Kubernetes Secrets committed in YAML
- Hardcoded credentials in CI variables
- Static database passwords — never rotated
✓ After
- KV v2 engine + dynamic database secrets
- AppRole & Kubernetes auth for pipelines and pods
- Short-lived credentials, auto-rotated on schedule
↑100%dynamic secrets
↓0hardcoded creds
↑3secrets engines
HashiCorp Vault
AppRole
K8s Auth
Vault Agent
CI/CD Automation with Shift-Left Security
Deployment time cut from 30 min to 5 min — Snyk + SonarQube + Trivy gates eliminated manual release errors.
✗ Before
- Manual multi-step deploys — slow, error-prone
- Vulnerabilities caught late — in production
- No quality gate between commit and release
✓ After
- 7 gated stages from PR validation to prod
- Snyk + SonarQube + Trivy scan every PR
- One-click image promotion via JFrog Artifactory
↑6×deploy speed
↓35%deploy failures
✓0criticals shipped
Jenkins
GitHub Actions
Snyk
SonarQube · Trivy
RHOCP 4.x High-Availability Microservices Platform
Multi-stage Helm promotion across DEV → PROD — 15+ microservices with instant rollback and zero-downtime deploys.
✗ Before
- Tribal-knowledge deploys, no rollback path
- Environment drift — dev ≠ staging ≠ prod
- 15+ devs pushing to one shared release branch
✓ After
- Helm multi-stage promotion with env values
- Versioned charts + instant rollback strategy
- Git branching strategy + runbooks in Confluence
↑15+microservices live
✓0downtime deploys
↑15+devs unblocked
RHOCP 4.x
Helm
Git Strategy
JFrog
Prometheus + Grafana + Fluentd Observability Stack
Proactive alerts fire before users notice — 10+ outages averted, logs and metrics finally correlated in one place.
✗ Before
- Reactive firefighting — alerts after users complain
- Logs and metrics trapped in separate silos
- Slow incident detection and diagnosis
✓ After
- Prometheus + Grafana dashboards with SLOs
- Fluentd DaemonSet → Kibana log-metric correlation
- Proactive alerts firing on anomalies early
↓40%MTTD
↓35%MTTR
↑10+outages averted
Prometheus
Grafana
Fluentd
Kibana
Terraform AWS EKS Infrastructure & Autoscaling
Slashed the AWS bill 30% with Terraform IaC + Lambda off-hours scheduling — 20+ services autoscaled through peak DTH traffic.
✗ Before
- Manual AWS console provisioning — slow, drift-prone
- Idle EC2 running 24/7 at full cost
- Over-provisioned capacity for peak traffic
✓ After
- Terraform IaC — EKS, VPC, EBS from modules
- Lambda scheduler shuts non-prod off-hours
- HPA + Cluster Autoscaler scale on real load
↓30%AWS spend
↑20+services autoscaled
↓60%manual toil
Terraform
AWS EKS
HPA
Ansible
