Jibu Chacko

Designed Kubernetes multi-cluster production architecture for GPU-backed inference and training workloads, optimizing for fault-domain isolation, workload segmentation, and cost-aware autoscaling strategies. Evaluated build-vs-buy decisions across provisioning, automation, and configuration management layers, defining long-term platform sustainability for very large node counts. Co-designed hardware/software acceptance and validation tooling with NVIDIA for critical inference and training reliability, and improved high-performance fabric management using NVIDIA InfiniBand, UFM, and Wekaclusters. Owned multi-year technical roadmap focused on hybrid bare-metal performance and cloud-native developer workflows, and established reliability governance (SLOs/SLIs, error budgets, observability standards) with cross-team alignment. Led high-impact incident response during complex scaling events and system outages, influencing CI/CD and dependency standards to prevent recurrence.

Introduced and implemented a Site Reliability Framework for a large and distributed Kafka farm. Contributed to platform-level architectural decisions impacting distributed Kafka infrastructure across multiple production environments while balancing resiliency, throughput, and operational complexity. Developed and managed platforms for developer experience on AWS and Kubernetes. Led development of a Go-based orchestration system governing lifecycle management of Kafka bare-metal nodes, establishing architectural patterns adopted across platform teams and reducing restart windows at scale. Automated infrastructure provisioning via a custom Terraform module to reduce provisioning time, led incident response for high-severity issues with strong post-mortems and root-cause analysis, and mentored engineers. Led integration of automated CI/CD pipelines for microservices, reducing deployment time and achieving near-zero downtime during releases. Championed comprehensive monitoring/observabilit

Inherted and expanded an engineering team of five engineers, mixing developers and SREs, and built a complete platform engineering team for bare-metal provisioning and asset management. Owned architectural direction for multi-datacenter bare-metal provisioning platforms supporting large server counts, defined lifecycle management standards, failure recovery patterns, and long-term scalability roadmaps. Set SLI/SLOs and SRE practices, established production readiness standards and SRE governance models influencing platform reliability across multiple engineering organizations, and defined an end-to-end incident management process using SLI/error budget practice. Spearheaded robustness improvements that improved system uptime, leveraging real-time monitoring and automated alerting. Spearheaded observability engineering initiatives improving incident response time using data from real-time monitoring tools to ensure smooth service continuity across multi-datacenter environments.