AWS DevOps Telemetry Backend

I need a seasoned DevOps engineer to stand up and run the entire backend that powers our transport-tracking platform. The system has to ingest GPS data from roughly 300 buses every 10 seconds, which works out to about 2.5 million write events each day, so resilience and low-latency processing are critical.

Platform & core stack
• Cloud: we’ll build everything on AWS (EC2/ECS/EKS, VPC, IAM, S3, Route 53 ‑ whatever fits best).
• Messaging: Redis is my first choice for the real-time pub/sub layer, though I’m open to Kafka if you can justify the trade-offs.
• Monitoring: Prometheus for metrics, with dashboards in Grafana; CloudWatch can complement for AWS-native alerts and logs.

Key things I expect you to deliver
• A fully scripted, infrastructure-as-code deployment (Terraform or CloudFormation) spanning networking, autoscaling groups, security, and storage.
• A telemetry ingestion pipeline that reliably buffers and forwards GPS data to downstream consumers and long-term storage.
• Redis cluster configuration tuned for high throughput and fault tolerance.
• End-to-end observability: Prometheus exporters, alert rules, log aggregation, and curated Grafana boards.
• A CI/CD workflow (GitHub Actions, GitLab CI, or similar) that ships containerised services from branch to production with zero downtime.
• Documentation that lets any engineer reproduce or extend the stack.

Acceptance criteria
1. Load-test results showing the platform sustains 2.5 million messages/day with <1 sec end-to-end latency at 50 % CPU.
2. One-command bootstrap of the full environment in a fresh AWS account.
3. Green pipeline from commit to production accompanied by automated tests and rollout verification.

If this sounds like your domain, tell me how you’d architect the solution on AWS and any improvements you’d suggest.