⏱️ 2,000+ hours across Go, Python & Kubernetes operators · building since 2023
My Kubernetes cluster has better uptime than my sleep schedule.
Software engineering student at UET Peshawar. Founder of Cortexium, building AI runtime sandboxes and backend orchestration infrastructure. Co-organizer of CNCF Peshawar. I contribute Go patches to production Kubernetes operators and implement distributed systems at the storage and protocol layer.
- Founder of Cortexium (AI runtime isolation and backend orchestration)
- Co-organizer, CNCF Community Peshawar
- Contributor to
openeverest/openeverest(Go Kubernetes database operator) - B.Sc. Software Engineering, UET Peshawar
openeverest/openeverest · Kubernetes Operator / Database Lifecycle
| Commit | Description | Root cause |
|---|---|---|
fix(rbac) |
Check informer init and wrap watcher error | RBAC informer panicked on cold-start due to nil dereference. Watcher errors were untyped, making them invisible in reconcile logs |
fix(monitoring) |
Wrap dropped errors in monitoring instance handler | Monitoring reconcile handler discarded errors instead of returning them to controller-runtime, silently masking reconcile failures |
fix(restore) |
Allow configuring resource limits on restore job containers | Restore Jobs had no CPU or memory constraints. Pods were unschedulable in namespaces with tight resource quotas |
| Project | Contribution |
|---|---|
| oppia/oppia | Core platform stability, issue triage, and code review. Active contributor since 2026 |
| AOSSIE-Org/EduAid | Resolved broken PyTorch version constraints that blocked CI across environments |
| AOSSIE-Org/Resonate-Website | Removed un-cleared GSAP timeline instances from React component unmount, eliminating a recurring memory leak |
| Project | What it does |
|---|---|
| aether-core-orchestrator | Backend orchestration layer at Cortexium. Runtime lifecycle, task routing, execution state |
| 30days-WebResearch | Concurrent research engine. SSE streaming, multi-source scraping, error budget enforcement |
| medcare-ai | Medical AI application |
| AdMatrix.ai | AI-powered ad targeting and campaign intelligence |
| last30days-skill | 30-day engineering practice tracker |
Production-quality implementations of distributed infrastructure components. Each covers storage, concurrency, networking, and failure recovery.
If it doesn't page at 3 AM, you haven't built distributed systems.
| Project | System | Implementation depth |
|---|---|---|
| blob-storage | Object storage (S3/MinIO internals) | XOR erasure coding, 3× replication, SHA-256 chunk integrity |
| rate-limiter-cluster | Distributed rate limiting | Atomic Lua token-bucket scripts, no race conditions across Redis nodes |
| web-crawler | Scalable web crawler | Priority URL frontier, politeness delays, cryptographic content dedup |
| search-autocomplete | Typeahead at scale | In-memory Trie, map-reduce trending aggregation, sub-30ms via CDN edge |
| video-transcoder | HLS media pipeline | 10MB chunking, concurrent async workers, broker-tracked job state |
| chat-backbone | Real-time messaging | WebSocket connection hub, heartbeat pruning, offline message queues |
| notification-router | Async multi-channel delivery | Exponential backoff with jitter, dead-letter queues |
| distributed-cache | In-memory KV store | O(1) LRU/LFU eviction, stampede shield, consistent-hash sharding |
| analytics-ingestion | Event ingestion pipeline | Kafka partitioning, Redis spike buffer, batch flushing |
| distributed-id-generator | Snowflake ID generation | 4,194,304 IDs/ms across 1,024 workers, clock-skew safe |
Building the infrastructure layer. The part no one sees until it breaks.