Safiullah K.

I build production AI systems, not prototypes. My focus is backend architecture and applied AI: FastAPI services, RAG pipelines, and multi-agent systems built with LangGraph that hold up under real client workloads. Recent work includes Obsidian, a multi-agent LangGraph platform built for Morgan State University, and ConstructIQ, a blueprint intelligence pipeline using OpenAI Vision and Pinecone. Core technical focus: ▪ RAG and retrieval systems: vector search, semantic chunking, and retrieval strategies built to reduce hallucination and keep responses grounded in real data ▪ Multi-agent orchestration: LangGraph-based agent workflows, tool-calling, and structured extraction pipelines ▪ Backend architecture: async, high-throughput FastAPI services, type-safe APIs, and microservices built for production load ▪ Enterprise automation: replacing manual operational work with automated backend workers and reliable background processing (Redis, Celery) If you're evaluating an AI architecture, optimizing a retrieval pipeline, or need a legacy backend rebuilt on FastAPI, let's talk through the scope and I'll map out an execution plan.