Zainab Athar

PROFESSIONAL PROFILE Please update objective section. EDUCATION Bachelors of Software Engineering School of Electrical Engineering and Computer Science , Islamabad , 3.5 (2022) INTERNSHIP EXPERIENCE Skylabs AI 01-Jul-2025 - 31-Aug-2025 Developed an automated evaluation pipeline for C++ by designing a hybrid framework combining SWE-Factory and Multi-SWE Bench. Focused on building a SWE-Bench for C++ to enable accurate and domain-appropriate golden patch generation for GitHub repositories. Tested the pipeline on multiple repositories including simdjson and fmtlib. Zhejiang University, China 31-Jul-2025 - 31-Aug-2025 Worked on automated entity and relation extraction for constructing subject–predicate–object triples. Utilized vertical- domain Englishand Chinesedatasetsincluding CHIP-20 and ReTACRED. Designed and planned a domain-specific pipeline, experimenting with preprocessing strategies and model architectures to improve Knowledge Graph construction Ayla 01-Oct-2024 - 30-Nov-2024 Developed an AI-assisted product to support patients with mild cognitive impairment. Designed conversational workflows using Large Language Models to aid memory recall and daily task assistance, integrating domain-specific prompts for improved usability and patient engagement. OffensioX 01-Aug-2024 - 30-Sep-2024 Identified and mitigated web application vulnerabilities, with a focus on preventing cross-site request forgery (CSRF) attacks. Collaborated with the security team to test, analyze, and strengthen application defenses against common exploit patterns. DataQuartz 01-May-2024 - 23-Aug-2024 Contributed to the development of QuartzGPT by integrating Large Language Models with Flowise and LangChain. As- sisted in building prompt-chaining workflows, orchestrating LLM pipelines, and enhancing system performance for inter- active AI-driven applications. FINAL YEAR PROJECT AI-Assisted Forensic Analysis System for Decision Support The proposed system is designed to support legal experts by automating complex data analysis tasks, thereby reducing the risk of human error while ensuring that final decisions remain under human control and aligned with legal and ethical standards. Potential features include tampering detection in images and videos, fake news detection with citations and corrections, voice identification with confidence scores and transcriptions, and handwriting analysis to identify authorship. All outputs must be explainable, auditable,

Please update objective section.