Divya Sonthalia

Cloud Platform and Site Reliability Engineer with 3 years of production experience across regulated OpenStack (Kolla-Ansible) environments — spanning private-cloud architecture, high-availability distributed systems, performance engineering, security hardening, and complex incident recovery. Deep hands-on expertise across Nova, Neutron, OVN/OVS, Cinder, Ceph, Octavia, Barbican, HashiCorp Vault, RabbitMQ, Galera, and HAProxy. Turns multi-layer production incidents into evidence-based root causes, controlled remediation, and reusable automation.

OpenStack SDE I at Ola – Krutrim (2025-08 – Present)

• Delivered production and pre-production OpenStack 2026.1 private-cloud environments via Kolla-Ansible — OVN networking, Ceph storage, high-availability control-plane services, Vault-backed PKI/TLS, and Barbican integration with signed images — with full control-plane and data-plane validation.
• Implemented security hardening and audit evidence for a regulated OpenStack environment aligned to ~150 MeitY/PCI-DSS-readiness controls, covering TLS, secrets management, image trust, access control, and key lifecycle.
• Designed and delivered a two-node Teleport privileged-access platform (PostgreSQL streaming replication, Pacemaker/Corosync, stable VIP, mixed-vendor BMC fencing, RBAC, MFA, session auditing); validated controlled switchover, live fencing, and automatic recovery, then onboarded 18 OpenStack nodes via guarded serial Ansible automation.
• Built an Ansible-based GPU node provisioning pipeline (BMC/iLO, OS hardening, Cortex XDR, Zabbix, Node Exporter, NVIDIA stack) and delivered 25 GPU nodes to an external customer.
• Diagnosed a major Octavia load-balancer performance gap to a single-CPU HAProxy bottleneck; implemented flavor-aware CPU mapping and rebuilt the Amphora image, then validated ~7.5K HTTPS req/s, ~1K fresh TLS sessions/s, and a 35–40K concurrent-connection boundary via repeatable benchmarks.
• Stabilized a three-node converged cluster after recurring Nova/Cinder/Octavia failures by proving ~89% of host RAM was locked in unused hugepages; executed an HA-safe rolling remediation that restored ~1.84 TiB of memory and reconciled 59 active VMs without service disruption.
• Modernized compute capacity by integrating hardened Ubuntu 24.04 nodes into an Antelope cluster while preserving Nova/Placement identities — managing Python 3.12, Docker pinning, and mixed-host-OS risks through staged rollout and rollback controls.
• Prototyped a Phase-1 OpenStack operations platform (Go control plane, React/TypeScript frontend, PostgreSQL, NATS JetStream, Loki) with audited container workflows and automated health-rule logic — validated backend/frontend builds and agent-to-server snapshot delivery.
• Additional engineering contributions: NUMA-aware VM scheduling & CPU pinning · Neutron DHCP high-availability design · Cinder secure-delete microversion tooling · live-migration network diagnostics · fleet-wide automation across 300+ nodes (kernel security audits, sysctl tuning, Docker health reporting).

Packaged App Development Associate at Accenture (2023-09 – 2025-07)

• Provided L2/L3 infrastructure support across Cisco and Juniper environments, maintaining uptime SLAs and leading RCA on escalated network incidents.
• Automated infrastructure health monitoring and recurring operational checks using Nagios, NetBrain, Efficient IP, Cisco Prime, and Fortinet NAC.
• Authored infrastructure and operations reports on incident trends and system health to support data-driven decisions for the network team.

BSc (Hons.) in Data Science – Sri Sri University, Cuttack (2020 – 2023)