Akash Singh

DevOps Engineer at Sabi | Kubestronaut | SIH ’25 | LFX’25 LitmusChaos | GSoC’24 Keploy | LIFT ’25 | Lead at Point Blank | 7x Hackathon Winner | Ex- FInalRoundAI, Aspora, CloudSek

• SA
  DevOps and MLOps EngineerSabi2026 to now- Built and operated multi-region Kubernetes (EKS) platforms driving full dev→prod environment parity. - Designed a push-based GitOps delivery pipeline with ArgoCD, External Secrets Operator, and Reloader for instant, secure config propagation across clusters. - Managed GPU model-serving infrastructure for LLM/TTS inference on Kubernetes using vLLM, deploying fine-tuned model variants as independently scalable pods with runtime model selection. - Automated infrastructure end-to-end with Terraform (IaC) and gated CI/CD promotion workflows, enabling safe, approval-based production rollouts. - Implemented cloud observability using OpenTelemetry-based collection and centralized metrics/logging across services.
• FR
  DevOps EngineerFinal Round AI2025 to 2026• Led complete Infrastructure as Code transformation using Terraform across AWS, Azure, GCP, Alibaba Cloud, and Hetzner bare metal — managing 4 EKS clusters, AKS, and K3s with Atlantis for PR-based plan/apply workflows and S3-native state locking • Drove end-to-end ArgoCD migration across 6 clusters with SOPS-encrypted secrets, zero-drift verification, per-app approval gates, and Argo Rollouts blue-green deployment strategy for independent multi-region releases • Hardened production Kubernetes: Karpenter autoscaling, Pod Anti-Affinity for HA, StorageClass Retain policies, Falco runtime security, and Beszel monitoring — achieving 90% VM host stability improvement • Architected multi-region LiteLLM proxy (EU/AP/US) with in-cluster PostgreSQL StatefulSets, retiring 3 RDS instances, supporting Azure Claude, Vertex AI, Groq, Cloudflare Workers AI, and Alibaba models across the company • Built production AI OnCall Engineer (DreamOps) with PagerDuty Events v2, Kubernetes MCP server, LiteLLM model routing, Slack incident threading, and auto-generated incident reports • Deployed comprehensive CI/CD pipeline reducing release time from 4 hours to <1 hour with parallel stages, build caching, automated rollback, and centralized reusable GitHub workflows across all services • Built end-to-end observability stack: kube-prometheus-stack, Loki, Tempo, Pyroscope, and OpenTelemetry auto-instrumentation for Python/Node backends — achieving <1 hour monthly downtime for critical services • Executed security hardening including staged secret rotation, self-hosted SonarQube with PR-based analysis, centralized TLS cert management via cert-manager across 9 domains, and BSI advisory response for bare-metal clusters • Reduced infrastructure costs by 20%+ through Elasticsearch migration to bare metal (65.8M docs/26 GB), legacy VM cleanup ($9,700/yr savings), orphaned resource removal, NAT Gateway consolidation, and Azure VM rightsizing
• TL
  LFX Mentee @LitmusChaosThe Linux Foundation2025My project at LFX Mentorship 2025 with LitmusChaos focused on modernizing the chaos engineering ecosystem by transitioning from Litmus 1.0 to 3.0 architecture. By developing a comprehensive Go SDK from scratch and revamping the Chaos-CI-Lib, I enabled seamless programmatic interaction with LitmusChaos for developers and SREs. My contributions included building production-ready SDK interfaces for authentication, experiment management, and real-time monitoring, while refactoring the entire CI/CD integration to eliminate direct Kubernetes API calls in favor of a cleaner SDK-based approach. Additionally, I optimized GitHub Actions and GitLab CI templates to reduce setup complexity, ultimately contributing approximately 14k lines of code across 4 repositories to democratize chaos engineering practices in cloud-native environments.

• DS
  Bachelor of Engineering - BE, Computer ScienceDayananda Sagar College of Engineering2022 to 2026

• kubernetes
• open source software
• orchestration
• site reliability engineer
• system administrator
• volumes
• api objects
• cloud
• containers
• custom resource definitions
• helm
• ingress
• logging and troubleshooting
• scheduling
• security policies
• services
• application containers
• apt
• aws cloud security services
• centos

• LinkedIn spotlight

  Speaker at FossAsia 2026

  (opens LinkedIn in a new tab)
• LinkedIn spotlight

  Volunteer at Google Summer of Code 2026 Booth

  (opens LinkedIn in a new tab)