Ritesh Sonawane

Site Reliability Engineer Now

Moniepoint · Cards and payments team

Where I am today: site reliability engineering on the cards and payments team.

Site Reliability Engineer 2024–2026

Cloudraft · Remote

Production Kubernetes across 10+ client environments in AI/ML, telecom, financial services, and enterprise SaaS, including Fortune 500 deployments. DevOps lead on enterprise clients, production on-call with incident.io, and SOC 2 / ISO 27001 support.

• AI security & GPU infrastructure Owned AWS EKS and VPC architecture; built 3 cloud-agnostic Helm charts running 70+ microservices in fully air-gapped AWS, GCP, and Azure deployments, including an 80B-parameter vLLM model on NVIDIA H100. Moved AI red-teaming orchestration from Celery to Argo Workflows and NATS, cutting compute cost 91%; a Go container entrypoint took pod startup from 5 minutes to 30 seconds.
• Zero-downtime observability migration Moved 100 TB+ of InfluxDB history to Grafana Mimir with a custom Go conversion tool and no downtime; rewrote 1,000+ dashboards from InfluxQL to PromQL; re-architected single-tenant Mimir to multi-tenant, cutting compactor startup from hours to 5 minutes and compute and memory use by 30%.
• Observability & cost Prometheus, Grafana, the OpenTelemetry Operator, Loki, and Tempo across 4 production AKS clusters with RED metrics and service graphs. Found $36,000 a year of waste with Steampipe and Powerpipe; led Kubernetes upgrades to v1.31.
• Enterprise SaaS & onboarding Replicated as the customer onboarding portal, taking onboarding from days to about an hour; multi-channel GitHub Actions pipelines for unstable, stable, and nightly releases; 6+ production environments onboarded across EKS and GKE with customized Temporal Helm charts.
• Multi-cloud & BYOC platforms kubeadm clusters provisioned with Cluster API on AWS, GCP, and Azure; Velero, Cluster Autoscaler, and Node Problem Detector automated with ArgoCD ApplicationSets. Plus a Go Kubernetes job scheduler with credit-based billing and a ledger for AI workloads.
• Supply chain & access SBOM generation and Grype scanning wired into CI/CD with Slack alerts; on-prem OpenFGA on CloudNativePG for fine-grained RBAC; org-wide SAML SSO for ArgoCD across a Fortune 500 deployment; bash deploys modernized to GitHub Actions and ArgoCD Image Updater.

B.Tech, Electronics & Telecommunication Jun 2024

ENTC · Graduated

Finished my engineering degree in June 2024, after a year of interning in DevOps alongside it.

DevOps Engineer 2024

Cloudraft · Remote · Internship

• GPU cloud platform Built a production GPU cloud for a GPU infrastructure provider on bare-metal Kubernetes (kubeadm) with KubeVirt and NVIDIA GPUs. Wrote the Go VM-provisioning API with GPU-aware scheduling, an x-api-key auth middleware for vLLM endpoints via Traefik, and Cilium CNI for public VM IPs. 99.9%+ uptime for commercial workloads, delivered with GitHub Actions, ArgoCD, and Harbor.

DevOps Engineer 2023–2024

Makerble · Remote · Internship

Impact-management platform combining CRM, scheduling, surveys, and reporting.

• Cloud migration & cost Staging, pre-production, and production moved from AWS to Azure with Terraform for a 40% cost reduction, plus 17% more from request/limit and node-affinity tuning.
• Delivery & reliability Tekton CI/CD with Slack automation; Ingress controller, Rollbar deploy hooks, and Robusta for Kubernetes monitoring; Prometheus, Grafana, and an EFK logging stack across all clusters.
• Internal platform Self-hosted WireGuard VPN for internal access and Passbolt as a password manager.

Where I started Before Kubernetes

Full-stack web & mobile

Explored full-stack web and mobile development, and won my first international hackathon along the way.

Homelab Ongoing

A multi-node k3s cluster for testing cloud-native patterns before they reach production.