I'm Bala Dengale.
I architect agentic AI platforms
on enterprise-scale Kubernetes.
Lead Systems Engineer & Platform Architect at Visa. Over the last decade I've architected and scaled Visa's global automation and container platforms — 140+ Kubernetes clusters across 10+ datacenters, with GitOps, service mesh, platform security and cloud-native modernization. Most recently I founded Visa's agentic AI platform for infrastructure operations — bringing AI-assisted troubleshooting, automation and operational intelligence to production at enterprise scale.
Agentic operations for Kubernetes & cloud-native infrastructure
Conceived, architected and led an AI-powered platform engineering and agentic operations framework for Kubernetes and cloud-native infrastructure. It runs a multi-agent operational ecosystem that integrates Kubernetes, ServiceNow, Tetrate, Quay, CMDB, observability and operational tooling to deliver AI-assisted troubleshooting, incident response, root-cause analysis and operational intelligence.
It's not a demo — it's adopted in production, where the agents triage incidents, run live operational SOPs and validate cluster health while the humans drink their coffee.
Current status: the agents carry the pager. I carry the espresso. 🔁
-
🤖
Declarative agents on KubernetesKagent-native platform buildout — agents expressed, deployed and operated as first-class Kubernetes citizens.
-
🔌
Custom Kubernetes MCP serversModel Context Protocol servers exposing cluster state, metrics, logs and control plane directly to agents.
-
🧠
Kimi model fine-tuningDomain-specific fine-tuning of upstream Kimi models on internal large-scale infrastructure datasets.
-
📚
Enterprise RAG architectureRetrieval-augmented generation over internal operational knowledge, SOPs and documentation.
-
🚨
Autonomous ops agentsSWAT incident resolution, live operational SOPs and automated cluster health checks & validations.
What I work with
Agentic AI & Ops Agents
Kagent-native platform buildout on Kubernetes, custom MCP servers, and specialized agents for SWAT incident triage, site reliability and automated cluster-health SOPs.
Custom Models & RAG
End-to-end enterprise RAG architecture, plus domain-specific fine-tuning of upstream Kimi models on internal large-scale infrastructure datasets.
Kubernetes Infrastructure
Multi-datacenter, multi-cluster enterprise Kubernetes platform engineering — GitOps, observability and operations at scale.
DevOps & Automation
Declarative IaC, bare-metal automated flows and CI/CD pipelines at enterprise scale.
Earlier chapters: BMC BAO/BSA orchestration, Cisco datacenter networking, cloud brokerage across AWS & Azure, and virtualization — 18 years across the stack, three eras of buzzwords, one constant: automate the boring parts.
Latest from the blog
Long-form experiments between cluster upgrades — written by a human, occasionally fact-checked by agents I built.
All posts →