Platform · DevOps · SRE
I'm Aditya Gupta. I build the layer everything else runs on — Kubernetes platforms, OpenTelemetry pipelines, Terraform estates — and I keep production honest: evidence-driven RCAs, zero-downtime changes, alerts that only fire when they should.
Led an org-wide migration to OpenTelemetry: instrumented 12+ services across Python, Go, and Kotlin/Java, consolidated three legacy tools into one platform, and built a GitOps pipeline for collector config with staging→prod approval gates.
Shipped a governed database-access platform on EKS: short-lived IAM tokens instead of stored credentials, four-eyes approval on every production query, full session audit. Eliminated shared credentials org-wide; hardened the upstream fork and cut its CI from 60 to 12 minutes.
End-to-end audit of the company's AWS estate with evidence-first sizing decisions — 14-day utilization data before any change, one-command rollbacks, zero downtime across every execution. Identified 15–20% in savings and executed the first tranche.
Production EKS via Terraform (private endpoint, KMS, IRSA, 3-AZ) with HA ZooKeeper and External Secrets Operator; zero-downtime blue/green ↔ rolling strategy switching and a tested five-minute hotfix path in the internal deployment platform; change-management process adopted org-wide.
One of two engineers owning the entire AWS estate — 25+ production services — for an at-home services platform scaling toward 10K TPS. Observability, deploys, databases, cost, incident response: the whole layer.
Real-time voice-AI infrastructure for Tier-1 banking: multi-region Kubernetes (India, Canada, Oregon), HA OpenSIPS telephony, PostgreSQL 12→16 migrations, HITECH compliance hardening.
Production fundamentals at national scale: ELK in production for consumer apps (Best Debutant Award), one-click Ansible deployments (−80% deploy time), zero-drama server migrations.
click any chip above for its story · shuffle for a random one
Prepping CKA (Kubernetes v1.35 curriculum) · learning Go to build a Terraform provider for Kviklet · writing up what migrating 20 services to OpenTelemetry taught me about tail-sampling economics · open to conversations about platform & AI-infrastructure roles.