~/whoamihover a tool to see what I do with it
I run Kubernetes on AWS and Azure, codify it in Terraform, and write the Node.js on top.
99.99%uptime / prod 2025
40%cloud spend removed
4m05sci build, was 8m12s
5 yrseks · aks on-call
KubernetesEKS · AKS
Production clusters with autoscaling, sane RBAC, GitOps deploys and alerts that mean something.AWSprimary cloud
VPCs, EKS, RDS and IAM, designed so that cost, access and blast radius are all easy to audit.Azuresecond cloud
AKS and managed services kept in step with AWS through shared modules and one review process.Terraformeverything as code
Every resource is reviewed in a PR and reproducible from scratch. Nothing gets clicked into being.Node.jsbackend
APIs and workers built to be observable from day one, so the platform can actually run them.CURRENTLY BUILDINGK8LensA clearer window into your clusters
ALSO IN THE TOOLBOX
AWS EC2EKSS3SQSLambdaCloudFrontEventBridgeAzure AKSAzure CDNCloudflareDockerKubernetesHelmHPANginx IngressTraefikMulti-arch buildsSpot InstancesTerraformeksctlJenkinsGitHub ActionsAzure DevOpsDocker layer cachingNode.jsExpressRedisRabbitMQPostgreSQLMongoDBMySQLCloudWatchCloudTrailPrometheusGrafanaSentryUptimeKumaIAM (least-privilege)VPCSecurity GroupsAzure VNetNSGsSSL/TLSACM
~/projectshover a card to open it
Selected projects
010204
In Development
15+K8LensOpen Source Product
15+resource types · intelligent health scoring · multi-arch Docker
K8LensOpen Source Product · H1 2026
Zero-config Kubernetes cluster manager - real-time pod, service, ingress monitoring with health scoring algorithm and multi-arch support.KubernetesDockerNext.js
Completed02
AWS to Azure Cloud MigrationCloud Infrastructure
Least-downtime migration · 20% infrastructure cost reduction
03AWS to Azure Cloud MigrationCloud Infrastructure · Q1 2023
Least-downtime infrastructure migration from AWS (EKS, EC2, RDS, ECR) to Azure (AKS, PostgreSQL, Container Registry) with Terraform and Jenkins.AzureAWSKubernetes
Shipped03
₹70,000WhatsApp Messaging Platform (BSP)Internal Product
₹70,000per quarter saved · turnaround cut from 7+ days to 2–3 days
WhatsApp Messaging Platform (BSP)Internal Product · H1 2022
In-house WhatsApp Business Service Provider - campaign management, delivery tracking, webhook handling, and CRM synchronization.Node.jsAWSDocker
Open Source04
Traefik DashboardOpen Source Product
Read-only monitoring - zero risk of accidental config changes
Traefik DashboardOpen Source Product · Q4 2025
Reverse proxy management interface - real-time monitoring of services, routes, middleware, and SSL/TLS configuration on Traefik.Next.jsDocker
~/principlespick one to see it in practice
Infrastructure as code, not as tribal knowledge
Every cluster, pipeline and alert is reviewable, reproducible and documented — so the platform outlives whoever built it.
01 / 03Reproducible clustersEKS and AKS provisioned from reviewed modules — environments that can be destroyed and rebuilt on a Tuesday.
40% cloud spend removed
cluster.tf
module "eks_cluster" { source = "./modules/eks" name = "prod-use1" version = "1.30" node_groups = { spot = { capacity = "SPOT", max = 24 } ondemand = { capacity = "ON_DEMAND", max = 6 } } budget_alert_usd = 1800}
02 / 03Pipelines with opinionsJenkins and GitHub Actions doing build, scan, promote. Cached layers took a 8-minute build under 4.
CI build 8m12s → 4m05s
pipeline.yml
stages: - build - scan - deploy build: cache: docker-layer script: - docker buildx build --platform amd64,arm64 - trivy image --exit-code 1 $IMAGE # 8m12s → 4m05s after layer cache
03 / 03Services built to be observedNode.js microservices with real health and readiness semantics, Redis and RabbitMQ behind them.
99.99% uptime / prod 2025
service.js
const app = express(); app.get("/healthz", (_, res) => { res.json({ ok: pool.healthy, lag: replicaLag });}); // readiness gates the rollout, not the podapp.get("/readyz", async (_, res) => { await redis.ping(); res.sendStatus(200);});
~/experience5 roles · 2 companies · since 2020
Five roles, one obsession with uptime
Product AssociateCareerLabs Technologies · Dec 2020 — Dec 2021 · Bengaluru · On-site
Full-stack POCs for stakeholder demos · contributed to 3 product launches.Full-stack POCs3 product launches
Software Development Engineer IICareerLabs Technologies · Dec 2021 — Dec 2022 · Bengaluru · On-site
Established Docker-based container pipeline; built high-throughput URL shortener and in-house WhatsApp marketing platform.URL shortener · 100k+ redirectsWhatsApp platform · ~₹70k/quarter saved
Software Development Engineer IIIConfiance Bizsol Pvt Ltd · Dec 2022 — Jul 2023 · Remote
Resolved latency bottlenecks across two high-traffic services for a 20–30% response-time reduction, and built data-import pipelines against third-party accounting systems.20–30% faster responsesAccounting data imports
Software Development Engineer L2Confiance Bizsol Pvt Ltd · Jul 2023 — Apr 2025 · Ahmedabad · On-site
Led the EC2-to-EKS migration and architected the Node.js microservice suite (paywall, email, file management) at 100% uptime. Cut lower-priority cluster cost 30–40% with Spot and on-demand fallback.EC2 → EKS migration100% uptime30–40% lower cluster cost
Assistant Engineering ManagerConfiance Bizsol Pvt Ltd · Apr 2025 — Present · Ahmedabad · On-site
Leading cloud-native architecture standards across the SaaS suite: IaC-first deployment on EKS, structured reviews, mentorship, and an OCR pipeline that cut document processing from hours to minutes.IaC-first on EKSOCR: hours → minutesTeam mentorship
~/writingthings I keep seeing break
Writing
Sep 5Stop Using nodeSelector for Node PlacementAug 26Stop Using 0.0.0.0/0 in Your Security GroupsAug 25Stop Allowing PostgreSQL Connections to BalloonAug 10Stop Over-Provisioning Your K8s Requests and Limits
Kubernetes7 min read
Stop Using nodeSelector for Node PlacementStop hardcoding infrastructure layouts in your manifests. It creates brittle clusters that fail to scale when your instance types change.kuberneteseks
Read postAWS8 min read
Stop Using 0.0.0.0/0 in Your Security GroupsOpen-to-the-world rules are a lazy security shortcut that leads to inevitable data exfiltration. Here is how to lock down your VPC traffic effectively.AWSSecurity
Read postBackend9 min read
Stop Allowing PostgreSQL Connections to BalloonUnbounded connection pools are a silent killer; here is how to use PgBouncer to prevent your database from buckling under a sudden spike in traffic.postgresqlpgbouncer
Read postKubernetes8 min read
Stop Over-Provisioning Your K8s Requests and LimitsIf you aren't measuring actual container utilization, you’re burning 40% of your cloud budget on idle CPU and RAM. Here is how to stop the waste.~/contactopen to work