💼 Professional Experience

🤖 AI-Powered Automated RCA

Built an agentic AI system that performs end-to-end Root Cause Analysis autonomously — just provide a deployment name and time range, and the agent does the rest.

• Architected an Agentic AI pipeline using LangChain integrated with MCP (Model Context Protocol) servers to connect with Kubernetes, Prometheus, Elastic APM, and Coralogix.
• The agent autonomously fetches pod status, queries metrics, pulls APM traces, and correlates Kubernetes events to pinpoint the root cause — zero manual investigation needed.
• Reduced mean-time-to-identify (MTTI) from 45+ minutes to under 5 minutes for common failure patterns.
• Applied prompt engineering to fine-tune agent reasoning chains, ensuring accurate correlation across noisy data sources.

⚡ Streamlining SRE with AI

• Smart runbook execution: When an alert fires (e.g. "Pod CrashLoopBackOff"), the AI agent reads the runbook, checks logs and metrics, and suggests the fix — engineers just approve and it executes.
• Capacity planning before big events: Before a major sale (e.g. Coldplay concert, IPL final), the agent pulls past traffic patterns for similar event types, factors in current ticket inventory and sale velocity, and generates a scaling baseline — giving engineers a data-backed starting point instead of hours of manual spreadsheet analysis. Final numbers are still tuned by the team since every event's scale is unique.
• Faster incident triage: Using prompt engineering to build reusable AI workflows that auto-classify incidents, assess change risk, and draft post-incident summaries in minutes instead of hours.

🔄 Legacy Bamboo → Argo CD Migration

• Led the migration from legacy Atlassian Bamboo CI/CD to Argo CD for GitOps-native continuous delivery.
• Designed a phased rollout strategy — migrating 100+ pipelines with zero deployment downtime.
• Introduced ApplicationSets and Helm-based Argo CD patterns for standardized, self-service deployments across teams.

📐 Streamlining IaC — AWS to Terraform/Terragrunt

• Existing AWS infrastructure (VPCs, EKS clusters, RDS instances, S3 buckets, IAM roles) was created manually or via CloudFormation — no single source of truth and drift was hard to track.
• Used terraform import to bring existing resources under Terraform state, then layered Terragrunt on top for DRY configurations across multiple environments and accounts.
• Now every infra change goes through code review and a plan/apply pipeline — no more ClickOps or undocumented manual changes in the AWS console.

💰 ARM64 Migration for Cost Optimization

• Spearheaded migration of Kubernetes workloads from x86 to AWS Graviton (ARM64) instances.
• Rebuilt multi-arch container images, validated compatibility across services, and rolled out progressively using canary deployments.
• Achieved ~30% infrastructure cost reduction while maintaining identical performance benchmarks.

🛡️ Disaster Recovery & Real-Time Data Resilience

• Always-on critical data: Optimized the DR setup so that critical components — payments, inventory, booking state — have real-time data replication across regions. If a region goes down, users don't see stale data or failed transactions.
• RDS cross-region read replicas with near-zero lag: Tuned replication to keep lag under seconds for business-critical databases, so the failover site always has a current view of live bookings and transactions.
• Automated failover validation: Built health-check agents that continuously verify DR readiness — confirming data sync status, connection pool health, and cache warm-up state — instead of relying on quarterly manual drills alone.
• Business continuity focus: Shifted DR thinking from "recover the infra" to "keep the business running" — ensuring that even during partial failures, users can still browse, book, and pay without noticing a thing.

🏗️ Technical Leadership

• System design: Owning architecture decisions for services handling millions of daily transactions — from database sharding strategies to API gateway patterns.
• Growing the team: Mentoring senior engineers through hands-on pairing sessions on Kubernetes debugging, observability stack setup, and building their first AI agents.
• Championing AI-first SRE: Driving adoption of AI tools across the org — e.g. replacing manual post-incident reviews with AI-generated RCA reports that teams can iterate on.