I build and validate AWS infrastructure around deployment control, measurable reliability, access boundaries, and cost-aware lifecycle management. That approach comes from two years of SLA-driven production NOC operations at Tata Communications, where incident prioritization, restoration commitments, and customer communication were daily responsibilities.

Two operating principles shape the work in these repositories:

Measure what users experience. Reliability indicators are scoped to user-facing routes and exclude health, metrics, and load-generation traffic.

State what the evidence does not prove. Controlled validation runs publish their limitations alongside their results rather than presenting short tests as production-scale evidence.

Focus: Terraform-provisioned AWS foundations · Kubernetes delivery on EKS with Helm and Argo CD · Python automation · Prometheus and CloudWatch observability · incident response · failure and rollback validation

An EKS reliability-operations platform for service health, deployment history, incident timelines, MTTR, and request-based SLIs, modeled after workflows used in production support.

• Provisioned a 59-resource AWS foundation with Terraform, including a three-tier VPC, managed EKS, RDS PostgreSQL, ECR, Secrets Manager, IRSA, ALB ingress, and CloudWatch monitoring
• Implemented 6 Prometheus recording rules and multi-window burn-rate alerts against a 99.9% SLO; verified behavior through 5 promtool cases and 12 assertions
• Load-tested the Python/FastAPI backend with k6: HPA scaled 2 → 6 replicas across 18,819 requests with zero failed requests or pod restarts, then returned to baseline
• Published SHA-256-attested validation evidence and verified teardown across 14 service checks, explicitly recording one inconclusive tagging check rather than reporting it as passed

Tech: AWS · EKS · Terraform · Helm · Python/FastAPI · Prometheus · RDS PostgreSQL · ECR · IRSA · ALB · CloudWatch · k6 · GitHub Actions · Checkov · kubeconform

Evidence: Repository · Validation Run · SLI Rules · Burn-Rate Runbook · Architecture

An EKS delivery platform that keeps Git as the source of truth for promotion, drift correction, and rollback.

• Composed the AWS foundation as 15 reusable Terraform modules for VPC, EKS, ECR, and IAM, with a $25 monthly AWS Budget guardrail for the validation environment
• Reconciled 4 Argo CD Applications across dev, staging, prod, and observability namespaces with scoped RBAC, ResourceQuotas, and Network Policies
• Secured promotion through a four-gate GitHub Actions workflow that validates Git and ECR provenance, authenticates through branch-scoped OIDC without stored AWS credentials, and delivers changes through pull requests
• Validated drift self-healing and Git-revert recovery after an injected readiness failure on a live EKS cluster

Tech: AWS · EKS · Argo CD · Terraform · Helm · GitHub Actions · OIDC · ECR · Docker · Prometheus · Grafana · AWS Budgets · Kubernetes RBAC

Evidence: Repository · AWS Validation · Promotion Workflow · Rollback Demo · Drift Detection

A deployed Python/Lambda availability monitor with scheduled checks, state history, state-change alerting, and a CloudFront-hosted dashboard.

• Deployed a 53-resource serverless stack with Lambda, EventBridge, DynamoDB, API Gateway, SNS/SQS, S3, CloudFront, KMS, CloudWatch, X-Ray, and Terraform
• Modeled a latest-status access pattern that reads 10 current rows instead of scanning 86,400 retained records in the reference workload, holding dashboard-read cost constant as history grows
• Hardened target handling against SSRF and redirect-based time-of-check/time-of-use bypasses; added API throttling, KMS encryption, dead-letter queues, and point-in-time recovery
• Published 9 application-level CloudWatch metrics and validated stateful alerting through an outage drill with 1-second detection, 26-second dashboard visibility, and zero duplicate DOWN alerts

Tech: AWS · Python · Lambda · DynamoDB · API Gateway · EventBridge · SNS/SQS · S3 · CloudFront · KMS · X-Ray · CloudWatch · Terraform · GitHub Actions · Bandit · Checkov

Evidence: Live Dashboard · Repository · Failure Drill · Design Tradeoffs · Security Notes

An event-driven incident-response workflow in which CloudWatch and SNS invoke a Python Lambda that assigns deterministic severity, generates remediation-focused summaries through Bedrock, and falls back to structured notifications when model output is unavailable. Terraform provisions the workflow; GitHub Actions runs Python unit tests, Bandit, Checkov, and IaC validation.

July 2022 – July 2024 · Pune, India

• Led five-engineer shifts in a 24/7 carrier NOC, coordinating triage for 40+ daily incidents across 25–30 Tier-1 clients and prioritizing concurrent P1/P2 outages against 99.9% availability and four-hour restoration commitments
• Diagnosed BGP/MPLS routing failures, TCP/IP faults, DWDM impairments, and hardware failures across Juniper, Huawei, Ciena, and Alcatel infrastructure
• Directed incident restoration through ServiceNow, coordinating vendor TAC escalation, field dispatch, customer-premises testing, and client communication
• Produced post-incident RCAs and cross-shift handoffs that preserved diagnostic context and supported recurrence prevention; earned a Certificate of Excellence for resolving an escalated customer incident

• AWS Certified Solutions Architect – Associate — Amazon Web Services, July 2026
• MS, Computer Science — Indiana University Bloomington, May 2026
• B.Tech, Computer Science and Engineering — SRM Institute of Science and Technology, 2022