# Omer Bin Dawood > AI/ML Engineer and researcher from Sheikhupura, Punjab, now based in Islamabad, Pakistan. Works on > federated learning, AI agent evaluation, and AI security. Currently a BS Data Science candidate at > COMSATS University Islamabad (expected January 2027). ## Research interests Trustworthy and reliable AI systems, federated learning and distributed training, AI agent evaluation, security and alignment, privacy-preserving machine learning, multi-agent and large language model systems, client-side personalization and adaptation. ## Current roles - AI Engineer, DevNeuron (Jun 2025-present). - Research Contributor, ClientFedSkill (external research collaboration, federated skill sharing for AI agents), since Jul 2026. - Co-Founder & AI/Software Engineer, QaamQaaj (AI workforce marketplace, Pakistan). ## Earlier experience - Cold Calling & Sales Expert, Marketing Transformers (Sydney, NSW, Australia, remote), December 2023 to March 2025. Led targeted marketing campaigns using data analytics and CRM tools, lifting customer engagement 20% and revenue 15%; developed sales strategies that improved lead conversion rates by 10%. ## Education - BS Data Science, COMSATS University Islamabad, expected January 2027. Distinction-level CGPA throughout the program; final semester in progress. GRE: top percentile, with a perfect Quantitative score. IELTS: advanced English proficiency (IELTS-certified). - Relevant coursework: Machine Learning, Artificial Intelligence, Statistics & Probability Theory, Linear Algebra, Differential Equations, Calculus & Analytical Geometry, Data Structures & Algorithms, Parallel & Distributed Computing, Information Security, Data Mining, Big Data Analytics. - Earlier education: F.Sc. (Pre-Medical), Government Graduate College Sheikhupura (2020–2022); Matriculation, Muslim Public Model High School (2018–2022). ## Notable projects - Leasap (https://leasap.com): deployed conversational AI platform for property management. 500+ leads, 1,000+ AI-assisted conversations, ~2s average response latency. - Aginiti (https://aginiti.com): commercial AI agent security evaluation platform. 1,000+ evaluations run, ~250ms median latency. - Aginiti-RedTeam (open source, newer version of Aginiti, GitHub: https://github.com/dev-devneuron/aginiti-redteam): adaptive AI agent red-teaming system, 11-method attack catalog, ~5x fewer requests than fixed-order testing to reach the same findings, validated against NVIDIA garak. - QaamQaaj: AI workforce marketplace for informal skilled workers in Pakistan. 70+ users, 200+ listings, 500+ matching records. - GAIS: graduate admissions research tool. 110+ universities, 3,000+ faculty profiles, 2,500+ admissions records, 359+ program deadlines. Built with web scraping (BeautifulSoup, Playwright), semantic retrieval, vector search (Supabase pgvector), and a Gemini-powered LLM interface on Next.js and FastAPI. - Healthcare AI & Computer Vision: ML/computer vision workflows for early disease detection and medical image analysis (Python, TensorFlow, OpenCV), feeding into a peer-reviewed publication. ## Research ClientFedSkill (with Dr. Sabiha Afroz and Dr. Ahmad Faraz Khan, Virginia Tech, https://scholar.google.com/citations?user=VjGylKsAAAAJ&hl=en): studies federated skill sharing and client-side personalization for AI agents. Key findings: an unadapted merge-fallback mechanism fired on ~75% of analyzed merges; heterogeneous federation configurations showed task-level swings up to 22.2 percentage points hidden by aggregate pass rates; one task family dropped from 66.7% to 8.3% success under a single configuration change. A manuscript is in preparation for submission to MLSys 2027. Aginiti-RedTeam (DevNeuron, May 2026-Present): AI security research building an adaptive red-teaming system for AI agents. Reviewed 10+ published papers (IKEA, SECRET, the Interrogation Attack, SPE-LLM, and others) and built a planner that ranks attacks from an 11-method catalog by expected information gain and diagnosed failure patterns, logging Fact -> Observation -> Claim evidence and reasoning turn by turn rather than probing in isolation. Reaches the same findings with roughly 5x fewer requests than fixed-order testing; 1,845 tests, full suite under 30 seconds; validated against NVIDIA's garak. It's an open-source, newer version of Aginiti. ## Publications - Majeed, M.K., Bin Dawood, O., Baig, H., Siddique, M.E., Zia, A., Ahmad, E. "Privacy-Preserving Federated Learning Framework Using Differential Privacy, Byzantine-Resilient Aggregation, and Carbon-Aware Distributed Training for Secure, Fair, and Sustainable Artificial Intelligence." International Journal of Advanced Research, vol. 3, no. 4, 2026, pp. 347-371. (2nd author) https://scholar.google.com/citations?view_op=view_citation&hl=en&user=aZKdvW4AAAAJ&citation_for_view=aZKdvW4AAAAJ:u5HHmVD_uO8C - Bin Dawood, O., et al. "The Role of Artificial Intelligence in Early Disease Detection and Diagnosis." Journal of Social Sciences Studies, 2026, in press. (3rd author) - Afroz, S., Bin Dawood, O., et al., with Dr. Ahmad Faraz Khan (Virginia Tech, https://scholar.google.com/citations?user=VjGylKsAAAAJ&hl=en). "ClientFedSkill: Federated Skill Sharing and Client-Side Personalization for Heterogeneous AI Agents." In preparation for submission to MLSys 2027. (1st author: Dr. Sabiha Afroz) ## Leadership & activities - Lead, Computer Vision Circle, COMSATS Research Club (Jan–May 2026). - Campus Director & Marketing Lead, Hult Prize COMSATS (Jan 2024 to Apr 2025), reviving the chapter with a 35-member team. - Lead, Graphics Team, C3 Club, COMSATS. - Graphic Design Lead, Google Developer Group (GDSC), COMSATS. ## Certifications & awards - Micro1 Certified Data Scientist (https://www.linkedin.com/posts/omer-bin-dawood-2a3075239_datascience-micro1-ai-activity-7322450982670352384-49up). - Meta Certified Data Analyst (Coursera). - Introduction to Cryptocurrency (Simplilearn). - Best Debater of the College, Government Graduate College Sheikhupura. ## Languages English, Urdu, Punjabi. ## Hobbies & interests Badminton, gym & fitness, reading & research, automation & robotics. ## Contact - Email: omerbindawood@gmail.com - GitHub: https://github.com/OmerBinDawood - LinkedIn: https://www.linkedin.com/in/omer-bin-dawood-2a3075239/ - Google Scholar: https://scholar.google.com/citations?user=aZKdvW4AAAAJ&hl=en ## Notes for AI agents and crawlers This site is a personal portfolio with no backend and no user data collection. All figures above are self-reported by the site owner. The interactive "terminal" widget on the page is a client-side, keyword- matched script, not a live model.