Headline

DevOps Engineer | Azure · AKS · CI/CD · MLOps · IaC · Python · Prometheus | Bengaluru

About

AI Infrastructure & MLOps Engineer building and operating production AI systems on Azure — Kubernetes, Prometheus, Terraform, MLflow, GitLab CI/CD. Currently at Comnet Infotech, where I own the observability and reliability stack for a Microsoft Azure AI partner platform serving enterprise clients including Flipkart, with Prometheus auto-discovering 20+ production VMs via Azure Service Discovery and 30 custom alert rules running 24/7. My work sits at the intersection where DevOps and ML systems actually meet in production: containerised AI services on AKS, end-to-end MLOps pipelines from model training through Docker containerisation to Kubernetes deployment, CI/CD with automated Prometheus rule validation on every commit, and infrastructure-as-code with Terraform for reproducible cloud provisioning. I built the platform's self-monitoring layer specifically because I found that a Prometheus crash would silence all alerts — no one would know. Prior experience: cloud infrastructure ownership for AWS/Azure SMB clients, a 40-user AD-to-Entra ID migration, and production cost anomaly tracing — work that built the operational instincts I now apply to AI infrastructure at scale. Open to MLOps Engineer, AI Platform Engineer, DevOps Engineer, SRE, Platform Engineer roles at GCCs and product companies in Bengaluru. Google Cloud certified. IEEE published. If you're building AI infrastructure and need someone who has actually operated it in production — let's talk.

Location

Bengaluru

AI stack

cursor

hugging face

ollama

openai

Work experience

DevOps & AI Infrastructure Engineer at Comnet Infotech, Feb 2025 - Present · 1 yr 3 mos, Bengaluru, Microsoft Azure AI partner — HaloGenie IVR platform (Azure OpenAI + Azure Speech) serving Flipkart, enterprise BFSI clients, and Indian Railways 139 helpline. • Designed production MLOps observability on Azure: Prometheus with Azure Service Discovery (zero-touch VM auto-enrollment via tags), Grafana, AlertManager (Slack/PagerDuty routing), 20 watchdog rules for self-healing monitoring. New infra appears in monitoring within 5 min of provisioning. • Built CI/CD pipeline for monitoring-as-code: GitLab CI with promtool/amtool validation on every commit, auto-deploy on merge, automatic rollback on config reload failure. Every config file versioned in Git. • Deployed containerised AI workloads on AKS: wrote Deployment, Service, ConfigMap, Secret, Ingress manifests; configured NGINX Ingress + TLS via cert-manager; debugged CrashLoopBackOff (OOM kill); implemented HPA for call volume spikes. • Built Python FastAPI microservice for AI-augmented alert routing: routes to Google Chat and MS Teams, extended with Azure OpenAI to generate 2-sentence triage summaries per alert. • Authored 30 PromQL alert rules across 4 rule files including disk fill prediction, CPU steal detection, Nginx p99 latency, OOM kill detection. Every rule has a runbook URL. • Owned Azure AI service infra: managed API endpoints for Azure Speech (ASR) and Azure OpenAI (LLM routing + post-call analytics), configured Private Endpoints, Blob Storage pipeline for transcription.

Cloud Infrastructure Engineer at Comnet Infotech, Aug 2024 - Feb 2025 · 7 mos, Delhi, India, Delhi-based SMB IT services firm — AWS/Azure cloud support, IT AMC, and OEM-authorised Lenovo/Asus service partner for manufacturing, trading, and logistics clients. • Administered AWS and Azure environments for SMB clients with no internal IT team: EC2, EBS, S3, Route 53, IAM, CloudWatch, Cost Explorer on AWS; VMs, NSGs, Entra ID, Azure Backup, Azure Monitor, Storage Accounts, Microsoft 365 on Azure. Full operational ownership from provisioning to access control, billing hygiene, and incident response. • Executed a 40-user on-prem Active Directory to Entra ID migration across two late evenings, with every change reversible. Zero business-hours lockout across the migration window. • Traced a 4x AWS billing spike to a GPU-backed EC2 instance running 23 days post-task; used Cost Explorer to drill to the instance, snapshotted EBS volume before termination, standardised billing threshold alerts at 120% of prior-month spend. • Enforced IAM security hygiene: rotated an exposed AWS access key found hardcoded in a shared-drive script, replaced with EC2 instance profile granting scoped S3 permissions, introduced access-key audits as a standing review step. • Resolved hybrid infra failures: VLAN mismatches, DHCP/static IP conflicts, L2TP and OpenVPN failures, NSG priority overrides, full-disk server slowdowns, post-DNS-change mail breaks.

Product Design Engineering Intern at GE, Jan 2024 - Jul 2024 · 7 mos, Chennai, Tamil Nadu, India

Education

Bachelor of Engineering - BE, Electronics and Computer, Amrita Vishwa Vidyapeetham