Bridging Neural AI with Symbolic Formal Verification, Corrective RAG, and Cloud Resilience
I am a Cybersecurity researcher and Systems Engineer pursuing my B.S. in Cybersecurity at Sir Syed University of Engineering and Technology (SSUET).
My work sits at the intersection of Formal Methods, Cloud Security, and Deterministic AI Systems. Rather than relying on heuristic prompt wrappers, I engineer neuro-symbolic architectures that couple neural LLM generation with mathematical constraint solvers (Z3 SMT) to guarantee provably correct, hallucination-free execution.
- Lead Author of Sentinel-Mesh: Coupled LLM patch synthesis with Z3 SMT formal verification for AWS cloud remediation (Preprint DOI:
10.21203/rs.3.rs-10674271/v1, under review at IEEE Transactions on Cloud Computing). - Open-Source Benchmark Architect: Creator of
CloudFix-Bench(105 verified cloud security benchmarks, Zenodo DOI:10.5281/zenodo.20975067). - Verified Peer Reviewer: Invited reviewer for IEEE Access (5 verified reviews on Web of Science, ResearcherID:
QIV-1552-2026). - Production Systems Engineering: Built fault-tolerant 7-stage Python ETL state machines processing 28,000+ records (5.6x speedup) and optimized AWS infrastructure overhead by 34%.
βββ π‘οΈ sentinel-mesh # Neuro-symbolic framework coupling LLMs with Z3 SMT for cloud remediation
βββ π browser-forensics-reconstruction # Deterministic DFIR log timeline reconstruction engine (Gemini + Heuristics)
βββ π vektor-ats-diagnostics # High-performance Document-VQA semantic evaluation engine for enterprise HR
βββ π¨ aegis-realtime-voice-dispatch # Low-latency streaming vocal command analysis & geospatial emergency routing
βββ π° Financial-Profitability-Guardrail # Postgres state machine & DLQ monitor ensuring 100% financial data integrity
βββ βοΈ Cloud-Security-Audit-Platform # Event-driven AWS security audit pipeline against 50+ CIS benchmarks