Aamir Hannan

• Built Next.js 14 AI dashboard with SSR/SSG, dynamic data fetching via TanStack Query, and shadcn/ui
component library — reducing Time-to-Interactive by 40% and improving Core Web Vitals by 25% via code
splitting and lazy loading.
• Engineered Node.js / Express.js REST & GraphQL API layer with JWT-based auth, rate limiting, and
Postman-documented endpoints consumed by frontend and third-party integrations.
• Designed PostgreSQL schema for multi-tenant SaaS (users, projects, usage logs) with optimised indexing
and pgvector extension for AI-powered semantic search across user data.
• Architected fault-tolerant microservices on Docker + AWS ECS with Redis Streams and MongoDB Change
Streams for real-time event pipelines, achieving 99.9% uptime under peak load.
• Optimised API throughput using async queuing (BullMQ), response caching (Redis), and SSE streaming —
cutting average response latency by 70% (1.3s → 400ms) and supporting 20K+ concurrent users.
• Integrated LangChain + OpenAI GPT-4o into the platform backend — building RAG pipelines with FAISS /
pgvector, semantic chunking, and hybrid BM25 + dense retrieval, improving answer relevance by 60%.
• Deployed LangGraph multi-agent orchestration (Supervisor-Worker topology) enabling parallel tool-calling
agents (web search, code execution, DB lookups) to resolve complex queries with stateful session tracking.
• Built LLM evaluation harnesses with Chain-of-Thought scoring and automated regression tests integrated into
CI/CD to catch prompt regressions before production.