The Researcher
Research Areas
Biography
I build agentic AI systems that know their own limits. My work focuses on the gap between autonomous capability and institutional accountability — designing architectures where AI agents can be stopped, audited, and corrected when they behave unexpectedly. I'm a 4th-year IT student at the Islamic University of Madinah and have published 19 peer-reviewed papers and manuscripts on AI governance, adversarial robustness, and constrained multi-agent systems.
Currently pursuing a B.S. in Information Technology on a merit-based fully funded scholarship, Ali has authored or co-authored papers across IEEE conferences, Springer, PLOS ONE, and MDPI — spanning agentic architectures, digital twins, AI governance, and safety-critical deployment. His research is motivated by a conviction that powerful AI systems must be understandable, governable, and safe by design.
Education
B.S. Information Technology
Islamic University of Madinah
Saudi Arabia
- Merit-based fully funded scholarship (Competitive selection)
- Current Cumulative GPA: 4.31 / 5.00
- Undergraduate researcher in agentic AI systems and AI governance
Thesis Interest
Safety-aligned multi-agent orchestration under Byzantine constraints.
Research & Professional Experience
Research Intern
King Fahd University of Petroleum and Minerals (KFUPM)
- Developing and evaluating HiVA, a value-aware hierarchical multi-agent RL system for wildfire suppression using Cell2Fire, with robustness testing across 16 distribution-shift conditions, including California↔Saudi transfer, rotated wind, ignition, and asset-layout changes.
- Contributing to the agentic extension of VNVspec, a machine-readable AI V&V framework, by developing typed governance controls for prompt security, tool authorization, memory leakage, human oversight, and multi-agent coordination.
Undergraduate AI Researcher
Islamic University of Madinah
- Designed agentic AI systems with explicit safety constraints, restricted action spaces, and policy-aligned decision rules.
- Conducted failure-mode and misuse analyses for autonomous systems, focusing on adversarial inputs, unintended behaviors, and oversight gaps.
- Explored LLM-based components within agentic pipelines: prompt design, orchestration, and robustness considerations.
- Investigated governance mechanisms for preventing unsafe behaviors triggered by untrusted or adversarial data sources.
- Analyzed system trustworthiness in scenarios analogous to log analysis, automated triage, and decision support pipelines.
Summer Intern
Xgrid.co
Lahore, Punjab, Pakistan (Hybrid)
- Developed a governance-aware autonomous Canary Deployment Simulator in Python for staged, cross-region software rollouts, integrating deployment orchestration, health analysis, rollback, and operational safety controls.
- Implemented automated failure detection and rollback workflows, including configuration-drift validation, forced/manual rollback, failure injection, and recovery handling for degraded deployments.
- Designed thread-safe deployment coordination with synchronized cluster state, asynchronous abort handling, idempotent rollback protection, and concurrency-safe audit logging.
- Built a governance and approval layer capable of blocking unsafe deployment transitions, enforcing risk-based policies, and requiring simulated human approval at critical rollout stages.
- Implemented operational resilience features including point-in-time cluster snapshots, region quarantine, staged recovery planning, deterministic event replay, causal audit tracing, and observability metrics.
- Developed automated testing and CI/CD quality controls covering unit, integration, stress, formatting, linting, coverage, and release validation workflows.
- Applied systematic code review and stress testing to identify and resolve concurrency, rollback consistency, and event-lineage issues in the deployment system.
Research Intern
IEEE EMBS E-JUST SBC
- Engineered biomedical image processing pipelines for automated feature extraction, noise reduction, and enhancement in diagnostic imaging datasets.
- Implemented deep learning architectures for image segmentation and classification, optimizing preprocessing workflows to improve model reliability.
Technical Methods & Expertise
Agentic AI Systems
Designing constrained reasoning pipelines with explicit safety guardrails and policy alignment.
Safety Engineering
Analyzing failure modes and edge cases in autonomous decision-making systems.
AI Governance
Implementing blockchain-based auditing and immutable oversight for AI agents.
Technical Writing
Distilling complex research into peer-reviewed publications and technical reports.
Awards & Certifications
Certificate of Appreciation (Research Excellence)
ICETAS 2026
Recognized for high-quality research contribution and presentation on Agentic AI Governance.
2026
Mindware: Critical Thinking
University of Michigan (Coursera)
Advanced certification in cognitive biases, statistical reasoning, and scientific methodology.
2024
Computational Thinking
University of Michigan (Coursera)
Formal training in algorithmic problem-solving and abstraction techniques.
2024
Merit-Based Fully Funded Bachelor's Scholarship
Islamic University of Madinah
Awarded for exceptional academic performance and potential in computer science and information technology.
2023