Feb 2026 - Present · Hybrid, Riyadh, Saudi Arabia

• Automated high-volume media migration (~2,200 files, ~273 GB) from Google Drive to internal infrastructure, reducing ~1 week of manual effort to 12 hours. Built an idempotent Python pipeline with codec validation, adaptive retries, and resilient error handling under API rate limits
• Architected a contract-first CMS Backend service from scratch to replace an unstable third-party vendor, with a configurable dual-write mirroring system enabling per-domain isolation of upstream breaks. Reverse-engineered a 20-table PostgreSQL schema from live API responses, automated Zod schema decomposition (2,800+ lines → 25 domain files), and delivered the full CI/CD and auth foundation
• Built and ran idempotent push notification scripts targeting ~30K users via FCM with full delivery tracking and documentation (~17K delivered, ~12K unreachable). Executed multiple targeted notification campaigns with zero duplicates
• Scripted bulk user cleanup of 40K+ records from CleverTap with idempotent execution and audit documentation
• Spotted a recurring HTTP 500 in Radio caused by an internal podcast discovery service silently failing requests, caught while tuning Datadog alert baselines
• Led production error triage across Radio, analyzing error volume across time windows to separate real regressions from traffic-proportional noise. Traced the root cause to a content-type mismatch in the frontend and redirected accordingly, avoiding an unnecessary backend change
• Designed and documented the Radio CMS architecture to shift recurring operational tasks (notifications, content ops) away from engineering, reducing team toil

Nov 2025 - Feb 2026 · Remote, Riyadh, Saudi Arabia

• Delivered multiple query and endpoint optimizations, including an ~11× p95 improvement on a timeline query and a ~20× latency reduction on a critical feed endpoint, using query decomposition and targeted indexing
• Resolved systemic data integrity issues — eliminated recurring foreign key failures from auth/DB state drift and fixed a race condition in a critical endpoint with Redis locking. Acted as first responder during a backend outage and strengthened monitoring safeguards
• Re-architected Datadog anomaly monitoring to separate downtime from spike detection, reducing alert fatigue. Designed a state-driven dynamic ads pipeline and established a Measure → Improve → Measure performance optimization process with documentation and tracking templates