Looking for a job

Lavkush Kumar

@lavkushry

Message

I build GPU-enabled MLOps platforms that automate reliable model deployment and observability.

India

Message

What I'm looking for

I'm looking to build and automate reliable ML platforms, GPU-enabled Kubernetes infrastructure, model-serving systems, and observability workflows that reduce operational toil and improve production delivery.

I've built GPU-enabled Kubernetes and MLOps platforms at CGI India and HFCL Limited, automating model lifecycles, infrastructure provisioning, and production deployment workflows.

At HFCL, I provisioned a 50-node Kubernetes platform for 12+ production ML microservices, reduced cluster bootstrap time from two days to four hours, and improved Triton serving throughput 2.5x while keeping p99 latency below 80ms.

I use Python, Kubernetes, Terraform, Ansible, Jenkins, Docker, MLflow, Airflow, and observability tooling to eliminate operational toil—from validation gates and GitOps rollouts to model drift detection and centralized logging.

Experience

Work history, roles, and key accomplishments

Current

DevOps / MLOps Platform Engineer

Current

CGI India

Jun 2026 - Present (3 months)

Orchestrated containerized ML workloads on AWS EKS using Terraform and Ansible, provisioning GPU-enabled node pools that achieved 99.9% environment alignment. Engineered automated CI/CD pipelines using Jenkins and GitLab CI, integrating Trivy security scans and model validation checks to accelerate release cycles by 60%.

Terraform Ansible Kubernetes Jenkins Gitlab CI Trivy Kong Gateway ArgoCD Helm GitOps

Software Enginner

HFCL

Jul 2025 - Jun 2026 (11 months)

Orchestrated GPU-enabled Kubernetes (kubeadm) cluster provisioning using Ansible for CNI config and Helm-based
deployments, reducing bootstrap time from 2 days to 4 hours.
• Engineered a 50-node Kubernetes platform hosting 12+ production ML microservices, configuring namespace isolation,
RBAC, and Horizontal Pod Autoscaling (HPA) to maintain 99.95% uptime.

DevOps