Debangshu Nath

Software Engineer | Distributed Systems • DevOps • Kubernetes • ML InfrastructureI’m a DevOps engineer focused on building cloud-native systems, deployment platforms, and real-time operational tooling. My work primarily revolves around Kubernetes, GitOps workflows, backend services, AI systems, and distributed application infrastructure.Building autonomous UAV infrastructure and AI-powered systems at Auraa

Projects

Kubernetes-Native AI Deployment Platform

A GitOps-driven Kubernetes infrastructure repository managing platform workloads, ingress configuration, ArgoCD applications, and dynamically generated AI inference deployments.Built using a multi-node kubeadm cluster with ArgoCD App-of-Apps architecture, Kustomize overlays, ingress routing, namespace isolation, and automated reconciliation workflows.The repository acts as the declarative infrastructure source of truth for dynamically provisioned LLM inference workloads.

A Kubernetes-native deployment platform for dynamically provisioning GGUF-based LLM inference workloads using GitOps workflows and llama.cpp.The platform automatically generates Kubernetes manifests, updates GitOps repositories, triggers ArgoCD reconciliation, and exposes runtime inference APIs through a React + Express dashboard.Supports dynamic model deployment, runtime inference interaction, automated manifest generation, and Kubernetes-native service discovery.

Real-Time Screen Translation Overlay

Developed a real-time desktop OCR translation overlay for translating Chinese on-screen text into English.Implemented asynchronous OCR pipelines and temporal stabilization to eliminate UI lag and translation flickering.Built region-based overlay rendering with collision avoidance to prevent recursive OCR feedback.Optimized rendering and update cycles for smooth drag/resize interactions during live screen capture.

<//EOF//>