Investigations, systems, and experiments.

I worked on the context-routing and export layer for a portable agent-skill bundle used across Codex, Claude Code, Gemini CLI, Antigravity, and OpenCode. The project reached 1,200+ npm downloads in its first 30 days and treats verified facts, constraints, and evidence labels as first-class agent inputs.

I developed a fraud-detection agent stack with LangChain and Langfuse, then tuned recursion depth, token use, and sampling settings to balance accuracy, cost, and latency under a tight contest budget.

I designed and benchmarked a multi-model defense pipeline across network-flow classification, malware API-sequence analysis, unsupervised intrusion detection, and Bash command tactic attribution. Reported experiments include 98.1% malware accuracy, 97.5% network-flow accuracy, and 95% anomaly precision.

I investigated verifier edge cases and mapped bypass candidates to reproducible exploitability conditions with C/eBPF PoCs, Python/Scapy/raw-socket triggers, QEMU, GDB, bpftool, verifier logs, and XDP execution paths.

I built an investigation workflow where AI-generated output must pass evidence-grounded checks before acceptance. The project emphasizes hallucination resistance, adversarial prompt robustness, and reproducible validation logic for forensic tasks.

I designed a multithreaded Python CLI that integrates Hybrid Analysis, VirusTotal, and ANY.RUN, then mapped 12 Equivocal Software Behaviours from 60 MITRE ATT&CK Enterprise techniques. The thesis analyzed 36 SourceForge goodware binaries and was published on Zenodo.

I processed 233,035 SSH honeypot sessions, including 90,026 Base64-obfuscated payloads, then trained TF-IDF, Bag-of-Words, Random Forest, SVM, Doc2Vec, and BERT classifiers for 7 ATT&CK-derived tactic categories.

I implemented core cryptographic primitives and protocol exercises in C and Python, then stress-tested them through cryptanalysis and CTF-style break/fix challenges. It bridges formal concepts with hands-on attacker thinking.

I engineered a secure timeout and watchdog mechanism on NXP S32K3 with FreeRTOS, custom startup/linker code, UART, timer handling, MPU-oriented memory protection, and QEMU build/run/debug loops for a 5-person university team.

I designed and compared IPSec and TLS VPN topologies in GNS3 using Cisco routers and Dockerized endpoints, then verified IKE/ESP security associations, TLS handshakes, NAT traversal, SecureNAT assignment, and protocol trade-offs with Wireshark and SoftEther logs.

I processed Android GNSS measurements for pseudorange and spoofing analysis, then ran WiFi goodput experiments with iperf, tcpdump, and Wireshark across Ethernet, WiFi, and mixed scenarios with repeated min/max/average/std-dev reporting.

I developed a security-first full-stack forum with RBAC authorization, JWT-backed sessions, TOTP-based admin 2FA, salted password hashing, parameterized SQLite queries, anonymous comments, and React/Express routing aligned to OWASP guidance.

I co-built Budget HQ, a fintech-focused web platform created during the IEEE-HKN Global Hackathon where we secured 1st place worldwide. The delivery balanced rapid product iteration with secure full-stack engineering.