Ask me about my work, just speak.
Ready
Not sure what to ask?
Start the call and try saying:
Abdur Rehman Afzal is an AI Engineer and Senior Backend Python Engineer based in Lahore, Pakistan (open to relocation), with 7+ years of experience shipping production systems across 9 live domains. He specializes in production RAG (Retrieval-Augmented Generation) pipelines, agentic and multi-agent AI, LLM-powered automation, and computer vision, on a strong backend foundation in Django and FastAPI. This voice assistant is one of his hands-on AI Engineering projects — it answers as Abdur and grounds every reply in his real portfolio using retrieval-augmented generation.
AI: AI Engineering, RAG, Agentic AI & multi-agent systems (LangChain,
CrewAI), LLMs, GPT-4/OpenAI API, prompt engineering, embeddings, semantic search, vector
databases (ChromaDB, Pinecone), AI automation (n8n), voice agents (Pipecat), computer vision
(YOLOv8, PyTorch, OpenCV, OCR).
Backend: Python, Django, FastAPI, DRF, Django Channels, Celery, RabbitMQ,
WebSockets, GraphQL, gRPC, REST APIs.
Cloud & DevOps: AWS, Google Cloud Platform, Docker, Kubernetes, CI/CD,
GitHub Actions, Nginx, Linux.
Frontend: React, Next.js, Angular, TypeScript, TailwindCSS.
Databases: PostgreSQL, MySQL, MongoDB, Redis, Firestore, BigQuery.
Who is Abdur Rehman Afzal?
An AI Engineer and Senior Backend Python Engineer with 7+ years of experience, specializing in production RAG pipelines, agentic AI, LLM automation, and Python backends (Django, FastAPI).
What AI projects has he built?
A RAG real-estate chatbot (+60% lead qualification), a multi-agent document-triage system (~70% faster), AI features for AdsGency AI ($50M+ managed ad spend), an Emirates ID document AI pipeline for a UAE government client, and this real-time voice assistant.
How was this voice assistant built?
A cascaded, low-latency pipeline orchestrated with Pipecat: WebRTC audio → Silero VAD (turn-taking & interruption) → Deepgram (speech-to-text) → OpenAI gpt-4o-mini (which calls a search_portfolio RAG tool) → Deepgram Aura (text-to-speech). Every turn is traced to Langfuse with OpenTelemetry. It runs as a Dockerized FastAPI service behind HTTPS, deployed via a GitHub Actions CI/CD pipeline.
How can I contact or hire Abdur Rehman Afzal?
Message him on WhatsApp (+92 334 1500 888), email abdurrehmanafzal786@gmail.com, visit his portfolio at abdurrehmanafzal.cloud, or connect on LinkedIn. He is based in Lahore, Pakistan, and his first priority is AI Engineer and Forward Deployed Engineer roles in Saudi Arabia, then the wider GCC, Europe and Australia.