Responsibilities

• Provisioned K3s Kubernetes cluster on OpenStack (3 nodes) and managed AWS resources (ALB, EC2, S3, CloudFront) using Terraform with daily drift detection
• Built DevSecOps CI/CD pipelines (GitHub Actions) with Trivy; reduced 47 Trivy vulnerabilities through pipeline-enforced remediation; deployed observability stack (Prometheus, Grafana, Loki) and Wazuh SIEM with Telegram alerting across hybrid nodes.
• Implemented ArgoCD GitOps with manual sync policy and enforced immutable image tags in production overlays
• Bridged AWS and OpenStack networks via WireGuard VPN tunnel; exposed internal services through Cloudflare Tunnel without public IPs
• Deployed kube-prometheus-stack (Prometheus, Grafana, Alertmanager) with hybrid node scraping across K3s and AWS; Loki + Promtail for centralized log aggregation; Telegram alerting for critical and warning events
• Deployed Wazuh SIEM with 6 agents across all nodes, integrated AWS CloudTrail S3 logs, and wrote custom Telegram alert integration
• Automated infrastructure auditing with Ansible and dynamic inventory from Terraform outputs; configured double ProxyJump SSH for private AWS nodes

Architecture

Frontend:  User → Cloudflare DNS → CloudFront → S3 (static)
API:       User → ALB → EC2 Backend (HA x2) → WireGuard VPN
              → OpenStack K3s → Database / OCR Service
Internal:  User → Cloudflare Tunnel → K3s Ingress (OpenStack)
              → Grafana / Prometheus / ArgoCD / Wazuh

Responsibilities

• Designed an auditing framework to evaluate GKE cluster security based on CIS Benchmarks, covering 31 security recommendations across control plane, node, and workload configurations.
• Developing and integrating modular audit scripts
• Validating security configurations to identify misconfigurations
• Applying remediation steps to harden the cluster environment

Responsibilities

• Contributed to building a CI/CD/CE pipeline using GitHub Actions to automate Docker builds and evaluation triggers
• Managed GitOps-based continuous deployment with ArgoCD for Kubernetes workloads
• Integrated "LLM-as-a-Judge" framework with GPT-4 and RAGAS for quantitative scoring of faithfulness and safety
• Implemented monitoring and logging stack (Prometheus, Grafana, Loki) with custom dashboards for real-time metrics

Responsibilities

• Constructed an Edge-Cloud testbed (Edge, Cloud, Gateway) using multi-node Docker environments
• Deployed medical LLMs (PhoGPT, VinaLLaMA) using Ollama and unified API communication layers
• Modeled inference offloading as an MDP and trained RL agents (PPO/DQN) to optimize decisions
• Simulated real-world network constraints (latency, packet loss) using Linux tc and netem

Responsibilities

• Provisioned cloud infrastructure and networking on GCP
• Configured multi-node distributed Spark cluster
• Monitored system resources and optimized cluster configurations

Responsibilities

• Designed testing topologies using VMware Workstation and Docker/Kubernetes
• Built and managed virtualized lab environments with VMware Workstation for isolated network testing
• Wrote python script to auto-measure throughput, latency, and packet loss using iperf3 and netem
• Analyzed system logs and compared performance metrics