BUILD
TEST
DEPLOY
LIVE

Richard Bonsu

Platform Engineer — AWS · Kubernetes · Terraform · CI/CD · AI-Agentic Workflows

location — Chicago, IL availability — on-site · hybrid · remote current — Innovid, since Nov 2022
View LinkedIn → Get in touch
01

About

service: richard-bonsu role: platform-engineer status: operational

Platform engineer with ~6 years running production infrastructure — from Hyatt's enterprise AWS environment to Innovid's ~10-service EKS platform. Deep in AWS, Kubernetes, Terraform, and GitOps, delivering CI/CD and observability that developers run themselves. Builds and runs agentic workflows (Claude Code, custom skills, Jira MCP), PR-gated against production infrastructure across 30 repos.

02

Platform Architecture

How a change gets from a pull request to running in production — the platform I built and run at Innovid.

Platform architecture diagram Pull requests reviewed by an AI agent, gating Terraform modules that provision dev, staging, and prod EKS clusters. Each cluster runs ArgoCD, an EFK logging stack with S3 archival, and Prometheus alerting. A read-only Claude Code skill reads the EFK stack for diagnostics. GitHub — 30 repos pull requests opened CLAUDE CODE + JIRA MCP PR-review agent flags gaps & exposed secrets pre-merge TERRAFORM Reusable modules · per-env tfvars rewritten to remove config drift EKS — dev validation-first EKS — staging pre-prod parity LIVE EKS — prod ~10 services ArgoCD auto-sync + self-heal deploys EFK + S3 centralized logs, archived Prometheus CPU/memory alerting CLAUDE CODE — READ-ONLY log-diagnostics skill 20-min Kibana/kubectl dive → one command — — — AI-agent path ─── infra provisioning path
03

Experience

HEAD
nov 2022 → present
Platform Engineer
Innovid · Chicago, IL · Hybrid
  • Engineered log-diagnostics, a read-only Claude Code skill that cut a ~20-minute manual Kibana/kubectl root-cause dive down to one command
  • Shipped an AI PR-review agent (a customized Claude Code GitHub Action with a Jira MCP integration) across 30 repos, flagging requirement gaps and exposed secrets before merge
  • Provisioned EKS clusters via Terraform with autoscaling for ~10 services across dev/staging/prod, replacing manual cluster setup
  • Wired up a centralized EFK (Elasticsearch, Fluentd, Kibana) logging platform with S3 archival across ~10 services, replacing scattered per-service log access
  • Rolled out workloads through ArgoCD with auto-sync and self-heal enabled, eliminating manual deployment steps across the platform
  • Set up Prometheus alerting on memory/CPU thresholds across ~10 services, giving the team early visibility into resource pressure
PRIOR
sep 2020 → oct 2022
Platform Engineer
Hyatt Hotels Corporation · Chicago, IL · Hybrid
  • Automated a GitHub Actions CI/CD pipeline, cutting deploy time from ~30 min to ~5 min
  • Migrated logging off CloudWatch to an EFK stack, adding full-text search across services CloudWatch didn't support
  • Tuned Grafana alert thresholds to cut noisy, duplicate pages, giving the team clearer dashboard visibility into service health
  • Tightened IAM least-privilege policies across roles
04

Stack

AWS

EC2ECSECRAPI GatewayVPCIAMSecrets ManagerS3RDSLambdaEKS

Infra & Containers

LinuxDockerKubernetes

IaC & Delivery

TerraformGitHub ActionsArgoCD

Observability

PrometheusGrafanaCloudWatchElasticsearchFluentdKibana

AI / Agentic

Claude CodeAgent SkillsMCP ServersCLAUDE.md / AGENTS.mdAI-assisted incident triage

Scripting

BashPython
05

Education & Certifications

University of Education, Winneba
Bachelor's, Construction Technology Education
Ghana
Kumasi Technical University
Higher National Diploma, Building Technology
Ghana
CKAD
Certified Kubernetes Application Developer
certification