Kaif Ul Imaan

PROFESSIONAL PROFILE Final year Computer Science student at the National University of Sciences and Technology (NUST) , Islamabad (H-12), specializing in Artificial Intelligence with a strong academic foundation in machine learning, deep learning, and large language models. Experienced in end-to-end data analysis, computer vision applications, and modern retrieval-augmented generation (RAG) systems, leveraging tools such as PyTorch, Pandas, PySpark, and Databricks to derive actionable insights from complex datasets and drive informed decision making. Currently developing a final year project on automated disease detection from microscopic cell images, targeting cross-modality generalization across tissue histopathology and blood smear microscopy. The work combines an in-depth review of state-of-the-art AI models and medical imaging platforms with hands-on experimentation using Hugging Face based PyTorch architectures, and extends to end-to-end system design including GPU accelerated training, cloud deployment on AWS, FastAPI based backend inference services, temporary tile storage pipelines, and explainable AI outputs tailored for clinician-friendly diagnosis. Eager to apply both theoretical knowledge and practical skills to challenging problems in AI research and industry, while continuing to expand expertise in emerging technologies and real-world applications. EDUCATION Bachelors in Computer Sciences School of Electrical Engineering and Computer Sciences (SEECS), NUST, Pakistan. , Islamabad. , 2.93 (2026) INTERNSHIP EXPERIENCE Ministry of Planning, Development and Special Initiatives Pakistan 21-Jul-2025 - 01-Sep-2025 Designed and implemented a deep learning–based Signature Detection and Verification System, utilizing YOLO for automatic signature detection from PDF documents and Siamese Networks for signature authenticity verification, forming an end-to-end computer vision pipeline for localization and similarity matching. Gained hands-on experience with LangChain and LangGraph to develop Retrieval-Augmented Generation (RAG) systems, including exposure to modern and modular RAG architectures, and explored LLM tool integration and agentic workflows for reasoning over external knowledge sources and structured tools. Cyberisk 03-Jun-2024 - 30-Aug-2024 I worked on data analytics and ETL pipelines using Python libraries, managed cloud resources through the AWS Console, and performed data processing and analysis with PySpark on Databricks. FINAL YEAR PROJECT Automated Disease Detection Using Microscopic Cell Images for Cross-modality generalization across two microscopy domains (tissue histopathology and blood smear microscopy)Automated Disease Detection Using Microscopic Cell Images for Cross-modality generalization across two microscopy domains (tissue histopathology and blood smear microscopy) This project tackles the challenge of cross-modality generalization in medical microscopy, bridging two distinct diagnostic domains: histopathology and hematology, across both standard microscopic images and whole-slide images. We leveraged Hugging Face- based foundation models that were pretrained specifically on microscopy data to establish a strong starting point. Our approach unfolded in three phases. In Phase I, we applied self-supervised domain adaptation using a DINOv2 student-teacher framework, allowing the model to learn robust representations from our datasets without requiring labeled data. This step was crucial for building

Final year Computer Science student at the National University of Sciences and Technology (NUST) , Islamabad (H-12), specializing in Artificial Intelligence with a strong academic foundation in machine learning, deep learning, and large language models. Experienced in end-to-end data analysis, computer vision applications, and modern retrieval-augmented generation (RAG) systems, leveraging tools such as PyTorch, Pandas, PySpark, and Databricks to derive actionable insights from complex datasets and drive informed decision making. Currently developing a final year project on automated disease detection from microscopic cell images, targeting cross-modality generalization across tissue histopathology and blood smear microscopy. The work combines an in-depth review of state-of-the-art AI models and medical imaging platforms with hands-on experimentation using Hugging Face based PyTorch architectures, and extends to end-to-end system design including GPU accelerated training, cloud deployment on AWS, FastAPI based backend inference services, temporary tile storage pipelines, and explainable AI outputs tailored for clinician-friendly diagnosis. Eager to apply both theoretical knowledge and practical skills to challenging problems in AI research and industry, while continuing to expand expertise in emerging technologies and real-world applications.