At Nutanix, I owned 75%+ of a Go-based Valkey Kubernetes Operator for standalone and clustered deployments. I implemented CRD-driven bootstrap, scaling, TLS/auth, PVC expansion, and S3 backup and restore.
I migrated ~20 Redis clusters to Valkey across 70+ environments supporting 200+ services, using RedisShake synchronization, data-parity checks, and controlled cutover. I also built self-service Kubernetes provisioning modules for Nutanix Clutch, including a Helm migration module that converts Git-managed Kubernetes YAML into Helm charts.
I evaluated ArcadeDB as a Neo4j alternative, benchmarking read/write performance and building graph data backfill and load-testing workflows. In an ArcadeDB open-source contribution, I replaced full-type vertex scans with indexed property lookups, improving read throughput from 3.6 to 7.5 RPS at 250K VMs and reducing a representative query from 1.3 seconds to 13 milliseconds.
I'm developing AEGIS, a Go agent runtime with Kubernetes-native Agent and KnowledgeBase CRDs and a RAG ingestion and retrieval pipeline. I’ve also published practical guides on GraphRAG, vector-search blind spots, and Redis cluster migration on Kubernetes.
