Shahbaz Raza

About

Backend/data engineer focused on high-scale, event-driven AWS ingestion systems (SQS/Lambda/EKS) processing millions of events per day, with strong reliability patterns (idempotency, DLQs, bounded retries) and deep observability using Datadog distributed tracing. Has delivered Terraform/GitHub Actions CI/CD and improved secret rotation via Secrets Manager + IRSA, plus Glue-based ETL with schema-evolution handling and Postgres SQL optimization (including JSONB/GIN indexing). Candidate is currently living outside the US and states they do not have US work authorization.

Experience

Education

Key Strengths

• Built AWS event-driven ingestion processing millions of events/day (SQS -> EKS workers -> S3/Postgres)
• Production reliability engineering: JWT auth, least-privilege IAM, explicit timeouts, bounded retries with exponential backoff/jitter, fail-fast on non-retryable errors
• Idempotent processing and deduplication using message IDs and Postgres upsert/unique constraints; DLQ routing for poison messages
• Strong observability implementation with Datadog APM/logs/custom metrics and distributed tracing across API, workers, SQS, S3, and Postgres
• Reduced alert noise by managing tag cardinality and using rate-based thresholds; created actionable monitors (queue lag/age, DLQ depth, error rate, query latency) routed to Slack with runbooks
• Resolved peak-time Postgres timeout/throttling issues by preventing retry storms; turned sustained incidents into short latency bumps
• Delivered automated AWS deployments using Terraform + GitHub Actions (build/test, Docker image to ECR, Kubernetes rollouts/health checks)
• Improved secret rotation reliability by migrating from static env vars to AWS Secrets Manager + KMS with IRSA and controlled rollouts
• Glue ETL troubleshooting: detected schema/type change failures via monitoring, identified root cause in logs, implemented safe casting and quarantined malformed records
• SQL performance tuning expertise (EXPLAIN ANALYZE, join rewrites, composite + GIN indexes) to eliminate sequential scans and reduce query latency (example: ~1.2s query)