Ifthekhar Mohammed

Grey Parrot Systems

Clients: Belle Tire, GE HealthCare

• Owned production Kubernetes on AWS (EKS) across Dev, QA, and Prod: Terraform-managed VPC, private API, IRSA, node groups, Argo CD, and Helm GitOps.
• Stood up the same platform pattern on GCP: GKE, Artifact Registry, Cloud Storage, Cloud IAM / Workload Identity, Cloud DNS, and Cloud Monitoring so workloads were not locked to one cloud.
• Built GitHub Actions pipelines that build, scan, and publish images to ECR and Artifact Registry, then promote environments without long-lived cloud keys (OIDC).
• Wrote Terraform modules with remote state per environment for AWS and GCP networking, clusters, and add-ons so Dev/QA/Prod stay parallel.
• Ran kube-prometheus-stack, Grafana, and Fluent Bit into CloudWatch and Google Cloud Logging; dashboards for pod health, load-balancer traffic, and node/GPU saturation.
• Hardened the platform: CrowdStrike on the cluster, GuardDuty and Inspector on AWS, least-privilege GCP IAM, SSO, and Kubernetes RBAC.
• Cut cloud spend on both AWS and GCP: Savings Plans / committed use, log retention, rightsizing node pools, and catching idle GPU and NAT cost.
• Deployed self-hosted LLMs on Kubernetes (vLLM / Ollama) on GPU node pools so internal prompts and documents stay in the VPC instead of a public model API.
• Shipped an internal OpenAI-compatible inference API in front of those models: auth, rate limits, model version in Git, GitOps rollbacks, and Grafana on latency and errors.
• Connected internal assistants / RAG to that private endpoint, with image scanning, egress limits, and no secrets in prompts or ConfigMaps.
• Partnered with app teams so promotions are a Git change—Helm values, OIDC, and environment parity—not kubectl apply from a laptop.

Tata Consultancy Services, India

Clients: Euroports, Forth Port, Adani

• Worked with a team of 10 to migrate a product from on-premises to Kubernetes on cloud landing zones, with workloads portable across AWS, Azure, and GCP.
• Designed cloud foundations—compute, object and block storage, load balancing, and autoscaling—first on AWS (EC2, S3, EBS, ELB) using patterns that transfer to Azure VM / Blob / Load Balancer and GCP Compute / Cloud Storage.
• Progressed CI/CD from Jenkins to GitHub Actions and Azure DevOps, cutting manual deployment work by about 70% and adding tests and image builds to the path to production.
• Deployed Kubernetes for high availability, improving application reliability by about 30% with health probes, rolling updates, and environment isolation.
• Automated patch management with Ansible across Windows and Linux, reducing critical vulnerabilities by about 80%.
• Ran ELK and Prometheus for real-time logging and metrics, catching issues earlier and reducing repeat incidents by about 25%.
• Introduced Terraform for repeatable network and compute so lower environments no longer drifted from production by hand-built consoles.
• Standardized secrets and least-privilege identities for pipelines and workloads instead of shared accounts and long-lived keys.
• Supported hybrid operations: on-prem servers, cloud landing zones, and container platforms in the same release cadence.