Backend engineer specializing in distributed systems and LLM-integrated applications. Currently Senior Software Engineer at Nayatel — Pakistan's leading ISP — fast-track promoted twice in less than 24 months, building production backend infrastructure: microservices, real-time event pipelines, and AI-powered features serving thousands of concurrent users.
I don't just use AI tools. I build the backend infrastructure that makes AI products work in production: RAG pipelines, agentic runtimes, vector search, streaming APIs, and the observability layer that keeps them reliable.
🧠 Chronos — AI Business Assistant Platform
Multi-tenant RAG SaaS platform combining a PDF knowledge base (via pgvector) with dynamic tool calling for real-time attribute retrieval. Streaming chat interface built on the Vercel AI SDK. Resolved production-grade challenges: edge-runtime PDF parsing failures, AI SDK v4 streaming protocol mismatches, and silent Zod schema crashes in LLM tool execution.
Next.js pgvector Gemini Embeddings (004) Vercel AI SDK TypeScript Supabase
🦞 Tiny-OpenClaw — Agentic AI Runtime
Production-ready agentic framework implementing the ReAct (Reasoning + Acting) loop via the Anthropic Claude API. Hot-loadable skills: web browser automation via Playwright, long-term memory, datetime. Fully async I/O with Python's asyncio. Migrated memory/session persistence from JSON to Redis for concurrent multi-user access and horizontal scalability.
Python Claude API Redis Playwright asyncio Docker Compose
⚡ Sportz — Real-Time WebSocket Architecture
Solved the stateful WebSocket fan-out problem across horizontally scaled Node instances using Redis Pub/Sub: mutations publish to a broadcast channel; every instance subscribes and routes to its local connections. Scales to N nodes behind a load balancer with zero dropped broadcasts. Includes heartbeat zombie cleanup, Arcjet rate limiting, Zod validation, and backpressure protection.
Node.js Express WebSockets Redis Pub/Sub PostgreSQL Drizzle ORM Docker
Backend
Databases & Messaging
AI & LLMs
RAG Pipelines Agentic Workflows (ReAct) Tool Calling pgvector MCP n8n
Frontend
DevOps & Observability
Nayatel — Pakistan's Leading ISP Oct 2024 – Present
Senior Software Engineer [Fast-Track Promoted ×2]
↳ Nwatch camera monitoring revamp NestJS · RabbitMQ · Redis · NewRelic · ~65% uptime improvement
↳ NayaTV backend modernization Microservices · Separation of Concerns · AI Recommendation Engine
Software Engineer
↳ Centralized payment platform Monolith → Microservices · 3+ internal apps unified
↳ CRON scheduling service Replaced unsafe crontab company-wide · Node.js · PostgreSQL
Assistant Software Engineer
↳ NayaTV OTT platform 40% faster content delivery · Stream load balancing
↳ Internal operational tooling Log monitoring · Complaint resolution APIs (TTR → minutes)
- 🔭 AI Recommendation Engine — data-driven content personalization for a live OTT platform
- 🛠️ NayaTV Backend Modernization — separation-of-concerns refactor on a high-traffic streaming service
- 📖 Going deeper on LangGraph, RAGAS eval frameworks, and distributed tracing