Ajay Akilan

Site Reliability Engineer (DevOps), Instant Messaging Platform — Bangalore, India

Service owner for Zoom’s core instant-messaging microservices — backend APIs, real-time messaging, and the public edge — with end-to-end production responsibility for releases, incidents, capacity and cost. Scope grew from single-service operations to running fleet-wide infrastructure programmes.

Zero-downtime infrastructure migration

• Led the replatform of the public-facing edge layer across the global production fleet on two clouds, using phased traffic cutovers with automated verification and failover validation at each stage — no customer impact, and edge infrastructure cost on the largest cluster cut in half.
• Owned the move of the messaging microservices from virtual machines onto Kubernetes within a company-wide containerisation programme — built the infrastructure definitions and deployment pipelines with verification embedded, cutting deploy time from hours to under 3 minutes and rollback to under 2.
• Ran the cache-layer transition from Redis to Valkey as a documented, repeatable process adopted as the team standard, and retired the legacy edge stack through staged decommission with full DNS, certificate and monitoring cleanup.

Reliability engineering through software

• Built the automated release-verification framework for the messaging services — from shell and Ansible scripts to gates native to the deployment platform, validating every pod on every production release — and co-developed the parallelised deployment workflow that together cut the multi-cluster release cycle from ~7 hours to 52 minutes.
• Wrote the Python CLI for production DNS operations on a 100,000+ record zone: resolution-chain tracing with live health-check overlay, bulk weight changes behind a preview–confirm–apply flow, and per-change audit logging. Primary tooling for all traffic cutovers; packaged and distributed to DevOps teams across the company.
• Automated the TLS certificate lifecycle for the global edge — zero-downtime rotation validated on isolated pre-production nodes — and built Python utilities for multi-repository management, cloud automation and log analysis, removing an estimated 10–15 hours of manual work weekly.

Multi-region architecture and multi-cloud operations

• Ran the active-active rollout for the chat backend and edge services across 14 of 15 production clusters — dual-region live traffic with weighted DNS routing and failover integrated with an internal traffic-orchestration console — at near-zero recovery time.
• Built production clusters from the ground up on three clouds, including one region delivered end to end: infrastructure, five services, active-active enablement and edge migration. Earlier builds on segregated accounts met government data-residency and compliance requirements.
• Incident commander for the messaging estate, on call from the third month of tenure: hundreds of P0 and P1 events resolved alongside NOC, development and platform teams, each turned into monitoring, alerting or runbook improvement.

Additional scope

• Drove cost optimisation across owned services through instance right-sizing, storage-mode tuning and Kubernetes resource refinement — six-figure annual savings with per-cluster reductions of 29–77%; authored the ongoing optimisation roadmap.
• Took the messaging services through the company's observability platform migration — hundreds of dashboards, severity-tiered alerting tied to runbooks, and SLI/SLO definitions set with the development team.
• Took two newly launched services through to general availability as lead DevOps owner; mentored 3–4 engineers and served as the standing escalation point across India, US and China engineering teams.
• Executed ~90% of global production releases from India through staged waves, establishing the India team's full operational independence.