DevOps Engineer
Feb 2026 – present
Own the AWS infrastructure, delivery pipelines and edge security for a high-traffic consumer platform: Terraform across three environments, serverless workloads, WAF and bot protection, and safe production change.
- AWS
- Terraform
- Lambda
- API Gateway
- AWS WAF
- ElastiCache Redis
- ECR
- CloudFront
- Route 53
- CloudWatch
- Kafka (MSK)
- GitHub Actions
- Python
- requests an hour behind the WAF
- ~900k
- requests an hour behind the WAF
- environments, all in Terraform
- 3
- environments, all in Terraform
- fraud signal types tracked
- 11
- fraud signal types tracked
- stored AWS keys in CI
- 0
- stored AWS keys in CI
AWS and Terraform
- Wrote and maintain Terraform for integration, UAT and production, with isolated state per environment so a change in one never touches another.
- Provisioned the full AWS stack for new backend services built by the development team: container-based Lambda, ElastiCache Redis with TLS and auth, Secrets Manager, security groups, CloudWatch and optional NAT egress switched on per environment. Existing VPCs are reused rather than duplicated.
- Manage shared networking in the central infrastructure repo: internal load balancer target groups, host-based listener rules and private Route 53 records.
- Brought hand-made WAF web ACLs and other console-built resources under Terraform by importing them, then made all new changes as code.
- Consolidated container images from per-account ECR repos into a shared cross-account registry, with cross-account pull permissions for Lambda and a cutover with no downtime.
- Found that an old integration VPC was effectively dead and moved workloads to the live one.
Serverless
- Run Lambda workloads as container images on arm64 for lower cost per request.
- Moved internal APIs off the public internet to private API Gateway endpoints, locked down with resource policies.
- Fixed Lambda rejecting multi-arch image manifests in the build, and added smoke tests to every deploy.
- Connected serverless workloads to a managed Kafka cluster over VPC peering or NAT egress, depending on the environment.
Scaling and reliability
- Run Redis Multi-AZ in UAT so failover behaviour is tested before production.
- Keep UAT in step with production, so infrastructure and WAF changes are proven on real-shaped traffic before they ship.
- Protect the platform under load with WAF rate limiting on the most-hit flows (login, activation, account recovery), tuned from real traffic so spikes and abuse are absorbed at the edge instead of reaching the backend.
- Every production change follows the same routine: backups, reviewed dry-run diffs, lock-token and expected-state checks so nobody's console edits get overwritten, and before/after traffic comparisons.
Security and WAF
- Own AWS WAF for the platform's APIs and for several Amplify-hosted frontends behind CloudFront.
- Rolled out AWS managed rule sets for IP reputation, common exploits and known bad inputs, Bot Control on the most-abused login endpoint, and Account Takeover Protection on a separate portal's login.
- New rules go live in count mode first, then get tuned and switched to block, so no legitimate users are locked out.
- Restricted admin paths to the corporate VPN with host-header and query-string rules.
- Audited the live WAF against code and found unmanaged console changes and rules that existed nowhere in code.
- Found and fixed a case-sensitivity mismatch that meant a fraud rule had never matched half its URLs, and rules sharing a CloudWatch metric name that made their numbers impossible to tell apart.
- Reviewed findings from an external security assessment, separating real gaps from ones already covered by a layer the assessors couldn't see.
CI/CD
- GitHub Actions pipelines that build images, push them to the central ECR, deploy and run smoke tests.
- Keyless pipeline auth through GitHub OIDC roles, with no long-lived AWS credentials stored anywhere.
Observability and fraud detection
- WAF logging, CloudWatch alarms and dashboards, including log-based dashboards where metrics weren't available.
- Built a Python tool that pulls 11 fraud signal types from application and WAF logs, including credential-stuffing patterns, suspicious IPs, activation velocity, card verification failures and device fingerprint spread.
- Showed that the 10k-row cap in CloudWatch Logs Insights meant some queries could silently miss results, and proposed fixes: pagination, tighter filters and input from the data and fraud teams.
Cost
- Moved Bot Control to the end of the rule order and scoped it to specific endpoints, so the paid inspection runs on far less traffic.
- Flagged WAF log volume (about 2.8 TB stored) as a cost item, and weigh query cost and time in all log tooling.
Documentation
- Handover docs for every service I stand up: environments, images, deploys, Terraform, routing, logs and access.
- Evidence write-ups for every production change: before/after configs, test results and rollback steps.