Where I've worked

Dscale for Research and Development · Dubai, UAE

1. 01Engineered high-performance backend services in Golang, delivering 20+ new features per quarter while following clean, maintainable coding practices.
2. 02Structured services around clean architecture and proven design patterns — repository, dependency injection, API gateway, and interceptor middleware — for testable, decoupled code.
3. 03Designed and implemented gRPC services with Protocol Buffers and gRPC-Gateway for efficient inter-service communication and REST API compatibility.
4. 04Orchestrated deployments of microservices on Kubernetes and Docker, improving scalability and reducing downtime by 40%.
5. 05Helped build GitHub Actions pipelines that ship 20+ Go microservices to Azure AKS across multiple environments, building images to a container registry and rolling out with Helm and Kubernetes manifests.
6. 06Re-architected the deployment pipeline solo into a tag-based release flow: environment-scoped, version-tagged deploys that select exactly which services ship, with clean promotion across environments and simple rollbacks.
7. 07Kept the cluster efficient and resilient with per-service Horizontal Pod Autoscalers that scale on real CPU and memory load, right-sized after profiling resource hot spots with htop and cluster metrics.
8. 08Built an event-driven backbone on Kafka using the transactional outbox pattern for reliable, ordered event delivery, lifting system throughput by 35%.
9. 09Utilized Redis for caching and session management, reducing API response times by 45%.
10. 10Leveraged Supabase and PostgreSQL for scalable, secure data storage, cutting complex query retrieval time by 30%.
11. 11Applied CQRS to split read and write paths — PostgreSQL as the source of truth, ClickHouse as the analytics read model, kept in sync through Kafka event envelopes.
12. 12Redesigned the Algolia search pipeline end-to-end, reshaping how listings sync from PostgreSQL through Kafka into search indices for faster, more reliable discovery.
13. 13Led data-engineering work to clean and remodel the search data, clearing indexing and data-quality issues that had sat unresolved for five years.
14. 14Improved search speed and relevance by tuning indices, query builders, and the sync path, and wrote reporting scripts that turn raw analytics into operational and business reports.
15. 15Established centralized logging and distributed tracing with Grafana and Loki, improving error detection coverage by 35% and reducing incident resolution time by 50%.
16. 16Built AI-powered auto-form filling on Google Gemini, cutting manual input time by 60% and boosting completion rates by 25%.
17. 17Built transactional email and multi-channel notifications — email, SMS, and Firebase (FCM) mobile push — with templated, deduplicated delivery through a dedicated notification service and background workers.
18. 18Delivered full-stack features beyond the backend, building admin dashboards and internal tools in React and Next.js on a shared, documented design system.
19. 19Reinforced authentication security with a Fibonacci-based exponential wait-time lockout, lowering brute-force login attempts by 70%.
20. 20Unified authentication flows across Apple, Google, UAE Pass, and username/password logins, improving login success rates by 18%.
21. 21Enforced authentication on every protected endpoint through a centralized gRPC interceptor with per-service token validation, plus Authy OTP for two-factor on sensitive flows.
22. 22Designed role-based access control with action-level restrictions and per-API-key endpoint scoping, keeping every role and integration to least privilege.
23. 23Secured inbound webhooks with per-provider signature verification and handler-level idempotency to block replay attacks and double-processing of payment and order events.
24. 24Engineered for safety under failure: idempotent operations, exactly-once event delivery through the transactional outbox, and Temporal and River retries so payments, orders, and notifications never drop or double-process.
25. 25Integrated core banking APIs with Checkout.com and Lean, and added RIVER-based queuing for refunds and captures with zero data loss.
26. 26Modeled auction and payment lifecycles as Temporal-orchestrated sagas, with River handling durable background jobs, retries, and scheduled work.
27. 27Ran River, a Postgres-backed job queue, in production for ad expiry, auction close, order completion, and multi-channel notification delivery, with automatic retries and scheduled jobs.
28. 28Owned data-engineering across the search and analytics pipelines in production: migrating and backfilling indices, reshaping the data model, and automating the sync and reporting paths into ClickHouse and Algolia.
29. 29Designed the data-classification and category taxonomy behind the classifieds platform (motors, property, general goods), with category-specific schemas and analytics, plus the end-to-end solution designs and ADRs that guided delivery.
30. 30Produced architecture research and production risk and opportunity reviews, migration plans, and deep technical documentation (service catalogs, runbooks, ADRs) that captured decisions and onboarded engineers.