PROFILE

AI Engineer with 2+ years of experience in LLMs, RAG systems, and Computer Vision. Skilled in Python, FastAPI, and SQL, with hands-on work in building chatbots, retrieval pipelines, and real-time video analytics. Strong in developing production-ready AI applications using LangChain, vector databases, and modern ML tools.

SKILLS
  • •Programming: Python, SQL
  • •LLMs & GenAI: LangChain, LangGraph, Agent routing, Tool calling
  • •RAG: ChromaDB, FAISS, Embeddings, Retrieval pipelines
  • •Computer Vision: YOLO, OpenCV, MTCNN, face_recognition
  • •Databases: PostgreSQL, SQLite
  • •Tools: Git, Docker, LangSmith, Alembic
EXPERIENCE

Xloop Digital Services

AI Engineer
11/2022 – 05/2025 | Karachi
  • •Designed and developed LLM chatbots, multi-agent systems, and RAG pipelines using LangChain and LangGraph.
  • •Built SQL and document retrieval workflows using vector databases and embeddings.
  • •Created real-time Computer Vision systems using YOLO, MTCNN, clustering, and OpenCV.
  • •Developed FastAPI services for RTSP video streaming, JWT authentication, logging, and ML model inference.
  • •Prepared datasets, preprocessing steps, and monitoring dashboards.
  • •Delivered production-ready Streamlit/Gradio apps for demos and internal tools.
    • PROJECTS

    DataBot

    Multi-Agent Chatbot with RAG & SQL | Xloop Digital Services
  • •Built an LLM-powered agent that answers queries using SQL databases, document collections, and web search.
  • •Implemented LangGraph agent routing, ChromaDB vector search, embeddings, and multi-tool orchestration.
  • •Used Qwen3-30B and Gemini Flash for reasoning and SQL execution.
  • •Added conversation memory, real-time monitoring (LangSmith), and a Streamlit interface for interaction.

    • VisionX - AI Surveillance System

    Textile Sector | Xloop Digital Services
  • •Developed a real-time AI surveillance system with 3 main use cases - crowd monitoring, violence detection, and face recognition.
  • •Crowd monitoring: Detected density levels (low → critical) using MeanShift clustering to track crowd behavior.
  • •Violence detection: Detected fights via CCTV feeds and triggered alerts on dashboard for specific locations.
  • •Face recognition: Tracked employee shifts and attendance in textile factories.
  • •Integrated FastAPI backend, SQL database logging, RTSP streaming, and dashboards for live monitoring.
  • •Labeled 12k+ images for training and system evaluation.
    • EDUCATION

    Dawood University of Engineering and Technology

    B.E. Computer System Engineering
    11/2018 – 11/2022 | Karachi, PK
  • •CGPA: 3.71/4.00