Hi! it's Amaan

AI & Backend Engineer — LLM agents, RAG, and production backends that hold up outside a demo

What I Do?

I build LLM systems that hold up outside a demo — agents with rules in code, retrieval that degrades gracefully, and traces you can debug. Software Engineer at CodingCops.🤝
My stack is as follows:

LLM Systems & Agents

⚡ Agentic workflows with LangGraph — intent routing, specialist nodes, and policy-as-code guardrails

⚡ RAG pipelines with Qdrant + SQL, conditional retrieval, and graceful degradation

⚡ Production FastAPI services with JWT auth, tracing (Langfuse), and evals (RAGAS)

⚡ Self-hosted inference (Qwen3 35B) for data residency on private GPU

⚡ Vue 3 chat UIs wired to real negotiation / agent backends — not demo toys

Backend & Data

⚡ Multi-tenant SaaS backends (Rails / Vue / Sidekiq) — full year at CodingCops before the AI work

⚡ PostgreSQL + Redis, RDS read replicas, and caching for heavy read paths

⚡ Vector search with Qdrant alongside relational grounding for agent turns

⚡ Directory sync workers (Azure AD / Google Workspace) and inbound email → tickets

⚡ AWS pieces in production: Lambda link previews, S3 ingest paths

How I Work

⚡ Prefer evidence over claims — traces, evals, and numbers from real systems

⚡ Ship end to end: ingestion → retrieval → agent → UI, not slides

⚡ Based in Lahore; open to AI and backend roles, remote or on-site

⚡ Reach me via LinkedIn or email — happy to walk through a public repo or architecture