Affaf

AI Engineer

Self Employed • Full time

Oct 2025 - Present • 8 mos

1)Developed ChronoScript AI, an AI-powered legal document analysis platform using FastAPI, PostgreSQL (pgvector), Celery, and Docker, with AWS S3 for storage and AWS Textract + PyMuPDF for OCR across PDFs, Word, and emails. Built a LangGraph/LangChain multi-LLM pipeline (Gemini 2.5-Flash) for automated timeline extraction, document classification, AI summarization, and entity recognition, orchestrated via a dual-queue async task architecture. Delivered end-to-end RAG with VoyageAI embeddings and pgvector for contextual case-level Q&A, with Auth0 multi-tenant auth, LangSmith tracing, Alembic migrations, and Stripe integration for subscription billing. 2) Developed, with my team, a full Arabic RAG-based ChatGPT system,named ShamelaGPT using LangGraph, FastAPI, PostgreSQL, and Milvus, starting from raw dataset acquisition. Scraped and cleaned millions of Arabic texts, performed preprocessing, deduplication, chunking, embedding generation, vector indexing, and orchestration of multi-LLM pipelines. Implemented language detection, query planning, retrieval, generation, and persistent chat history, enabling highly contextual QA across classical Islamic literature. ● Co-developed MPI Plumbing ERP Fast API backend, enhancing its AI-powered chatbot system capable of natural-language database queries, automated Excel generation, email dispatch, PDF-to-Excel invoice parsing, supplier tracking, and QuickBooks export logic. Improved multi-agent communication (Database, Excel, Email agents) and optimized API routes, data models, and background task execution to deliver faster and more accurate pipeline performance across the system.