Himaya AI
Associate Member of Technical Staff,… at Himaya AI · India
Entry LevelRemoteFull-timeIndia268000 INR–287000 INR/yr
Job Description
About the role
You will keep Himaya's production footprint reliable and secure across two clouds, and you will build the CI/CD that the team currently lacks. Today deploys are scripted ACR builds pushed to Azure Container Apps by hand. Your first quarter is making that a pipeline with tests, image scanning, and staged rollout.
What you will do
- Own the Azure production estate: Container Apps revisions and traffic splits, ACR, Azure Database for PostgreSQL, Service Bus, Key Vault, Front Door, and the Azure OpenAI deployment.
- Own the AWS estate: ECS services, RDS, ElastiCache, SQS queues with DLQs, S3, CloudFront, and the isolated EC2 detonation-sandbox VPC.
- Build CI/CD on GitHub Actions with OIDC workload identity (no long-lived secrets), running the pytest suite, tsc, and production builds as merge gates, then deploying by revision with health checks and rollback.
- Secure the microservices and codebase: scope service principals and IAM roles to least privilege, rotate and centralise secrets, harden container images, and burn down the dependency-vulnerability backlog with a repeatable process.
- Improve observability: Sentry releases and environments per service, structured JSON logging, alerting on queue depth, scan-loop failures, and revision health.
Requirements
- 1 to 3 years in infrastructure, DevOps, or backend work with real production responsibility.
- Hands-on with at least one major cloud (we run Azure and AWS; you will learn the other here).
- You have written a CI/CD pipeline from scratch, GitHub Actions or Azure DevOps, and can explain OIDC federation versus stored credentials.
- Comfortable with Docker, container registries, and debugging a failed deploy from logs alone.
- You treat IAM as a security surface: least privilege, scoped roles, no shared admin credentials.
- Nice to have: Postgres operations (backups, connection limits, lock contention), Terraform or Bicep, experience running queue-based workers, exposure to compliance regimes (SOC 2, UAE or KSA data residency).