Chaitanya Sistla

symptom

After an EKS AMI upgrade, ingester pods began crashing and ingestion stalled.

diagnosis

Traced the failure to the node image change introduced by the AMI upgrade rather than the application itself.

fix

Diagnosed and resolved the crash so ingesters returned to healthy operation.

lesson

Node-level changes are application-level incidents. Treat AMI upgrades as first-class changes with the same scrutiny as app deploys.

symptom

Requests returned 404s where routes were expected to resolve.

diagnosis

Identified missing HTTPRoutes as the cause of the failed routing.

fix

Restored the missing HTTPRoutes and traffic resolved correctly.

lesson

Gateway routing config is part of the deploy surface. Missing route objects fail quietly as 404s, so verify routes as part of rollout.

symptom

The query tier went down, cutting off read access to telemetry.

diagnosis

Led the investigation into the query tier failure under live outage conditions.

fix

Led resolution and brought the query tier back to service.

lesson

Read-path availability matters as much as ingestion. An outage on query is an outage for every dashboard and alert that depends on it.

symptom

More than 1200 servers carried known vulnerabilities, with configuration drifting under Chef.

diagnosis

Remediation had to be fleet-wide and fast. The Chef-managed config had drifted enough that per-host fixes would not hold.

fix

Converted the entire fleet from Chef to Ansible overnight, bringing configuration under one consistent, idempotent layer and removing the vulnerabilities across 1200+ servers.

lesson

A uniform, idempotent config layer is a security control. Fleet-wide remediation is only possible when every host is managed the same way.

symptom

A regulated banking platform ran Kafka on 25 physical servers provisioned by hand, outside any version control.

diagnosis

The fleet had to come under infrastructure as code, but nothing could be recreated: the servers were live and in production.

fix

Imported the existing physical servers into Terraform state so they were managed as code in place, then used Ansible for services management, with no recreation.

lesson

Terraform import turns brownfield physical infrastructure into managed code without a risky rebuild. Import first, converge second.

Founder and shareholder, non-operational

An AI-native preventive cloud security platform. The positioning was fix and govern, not find and alert. What I built:

• +Multi-cloud discovery pipeline: read-only scans of AWS, GCP, and Azure, normalized and enriched by 150+ provider enrichers, correlated into a relationship graph, matched against IaC inventory for managed-versus-orphan and drift.
• +Security scanning across three engines (Nuclei, Prowler, Cloud Custodian), correlated by resource and vulnerability class into confidence verdicts, mapped to CIS, NIST, SOC 2, PCI, HIPAA, and ISO.
• +Resource relationship graph: 100+ metadata-driven rules per provider, hard and soft dependencies, blast-radius analysis, and Kahn topological sort for safe Terraform import ordering.
• +Claude-powered analysis on the Anthropic SDK: grounded single-turn remediation and Terraform generation over a knowledge graph of findings, with dynamic guardrails, strict output validation, and per-call token and cost metering.

+ 3 more

Creator

An open-source Rust-based parallel infrastructure-as-code execution engine.

• +Parallel execution engine for infrastructure as code
• +Written in Rust

Creator

An open-source CLI, launched on Product Hunt.

• +Open-source command line tool
• +Launched on Product Hunt

Author and maintainer

The Terraform provider and Kubernetes modules for OpenObserve, published to the Terraform Registry.

• +Terraform provider for OpenObserve
• +Kubernetes modules
• +Published to the Terraform Registry