Sumbal Hanif

Research Artificial Intelligence Engineer with a strong foundation in deep learning and machine learning, experienced across diverse AI domains, including autonomous systems, deep learning & machine learning, computer vision & medical imaging, LLMs & generative AI, text- & graph-based data, with hands-on expertise in training models on NVIDIA and MacBook GPUs and skilled in Python, Django, Linux, Docker, and AI frameworks such as TensorFlow, PyTorch, and Keras, Hermes AI, and OpenClaw.

Artificial Intelligence Engineer | LLMs | RAG | NLP | Machine Learning | Knowledge Graphs | Python - Power Planning and Monitoring Company (Ministry of Energy) - Islamabad, Pakistan

(2025-08)

• Hermes AI: Collected and processed live Power-sector data, stored it in the database, and built a Qdrant-based RAG layer for knowledge-grounded retrieval.
• Integrated Hermes with the RAG knowledge base and enabled live A2A communication between agents for autonomous, visible multi-agent conversations.
• Implemented agent-to-agent debate where responses are exchanged, challenged, and synthesized in real time, with the conversation visible to the user.
• Knowledge Graph: Developed Knowledge Graph and Retrieval-Augmented Generation (RAG) pipelines for document intelligence using LLMs, embeddings, Neo4j, and Python.
• Built automated entity extraction, relationship extraction, and graph construction pipelines from PDF and DOCX documents using Python.
• Generated JSON-based Knowledge Graphs and integrated them with Neo4j for graph storage and querying.
• Designed interactive PyVis/HTML visualizations for exploring entities, concepts, and semantic relationships.
• Developed and optimized prompt engineering workflows for Large Language Models (LLMs) to automate document understanding, structured information extraction, and intelligent question answering.
• Virtual CEO: Contributing to a multi-agent "Virtual CEO" system that analyzes real-time data from news and social media to generate intelligent insights and business recommendations. Integrated data pipelines, AI (RAG), and decision layers (SIR & SAM) to automate strategic decision-making and alerts.
• PowerPost: Built an AI-powered energy news platform with Python and Apify-based web scraping pipelines, automated data processing, and PostgreSQL vector storage, enabling semantic retrieval and AI-generated summaries of global energy news.
• Energy Data Dictionary: Created an AI-powered dictionary defining energy terms, ranges, and technical concepts.
• AI-Based Document Extraction: Designed a prompt-based system to extract tables from complex PDF documents using AI.
• Planner Automation: Automated a task monitoring system using Python to replicate an n8n workflow, integrating PostgreSQL and Google Calendar API. Replaced a manual workflow tool with a fully coded scheduler that queries the database, summarizes tasks, and updates shared calendars automatically.
• Built AI APIs using Python to integrate AI models into production workflows and automated digital services.

Research Associate (Artificial Intelligence) - National Center for Cyber Security (NASTP) - Rawalpindi, Pakistan

(2025-06 - 2025-08)

• Fine-tune Mistral7b for multimodal cyber complaints analysis (text, voice, images, documents).
• Integrated Whisper for voice and Tesseract for OCR on images and documents.
• Collaborated with a team to handle cybercrime complaints.
• Prepared comprehensive documentation in SRS format to support structured system development.

Machine Learning Intern - BiStartX - Lahore, Pakistan

(2025-04 - 2025-05)

• Predicting House Prices Using Linear Regression
• Classification of Iris Species Using Logistic Regression
• Predicting Diabetes Using Decision Trees
• Predicting Employee Attrition Using Random Forest

Junior Artificial Intelligence Engineer - Xylexa Inc. (NSTP) - Islamabad, Pakistan

(2024-07 - 2024-09)

• Preprocessed CT and X-ray medical images.
• Prepared labeled datasets for model training.
• Fine-tuned models for medical imaging tasks.
• Trained and validated diagnostic models.
• Object Detection: Breast Cancer Detection Using YOLOv4
• Semantic Segmentation: Brain Tumor Prediction Using U-Net
• Generative AI: Fine-tune BioGPT and T5 for Text Summarization
• RAG and LangChain: BERT and T5 for ChatBot

Computer Science Lecturer - Govt. Viqar-Un-Nisa Post Graduate College - Rawalpindi, Pakistan

(2021-02 - 2022-04)

• Delivering lectures and stimulating instructional sessions on Artificial Intelligence
• Facilitating hands-on coding practice in the lab on C/C++

Master's of Philosophy (M.Phil.) - Computer Science - Quaid-i-Azam University Islamabad (2022 - 2024)

Bachelor's of Science - Information Technology (BSIT) - University Of Gujrat (2016 - 2020)