DNS Resolution Monitoring
CAA-record analysis for TLS-interception detection, plus a passive-DNS hunt for the SolarWinds/SUNBURST C2 domain.
Cloud Security & AI/ML for Cybersecurity
M.Tech in Cybersecurity Engineering, with hands-on work spanning AWS security, SIEM-based threat hunting, digital forensics, applied cryptography, web application security, and machine learning for malware and fraud detection.
Upload a photo or paste a product link, auto-detect the dominant color, tag the type, and search "blue top" later. Built and deployed solo, end to end — including provisioning the VM and cutting its TLS certificate.
Hands-on labs from an M.Tech Cybersecurity program, grouped by domain.
CAA-record analysis for TLS-interception detection, plus a passive-DNS hunt for the SolarWinds/SUNBURST C2 domain.
tshark-driven pcap analysis that surfaced a live phishing domain impersonating a major Chilean bank, confirmed against real query traffic.
Hunted brute-force logons, a port scan, and malware alerts by querying Wazuh's indexer API directly with curl.
Traced a malicious hash to the SystemBC backdoor trojan, then pivoted to Shodan to profile its C2 infrastructure.
Stood up a Kali attack VM and AWS account with scoped IAM access keys.
Built a segmented AWS network: public/private subnets across two AZs, a NAT gateway, and a default-deny NACL.
Worked the well-known flaws.cloud series: public S3 buckets, leaked git credentials, EC2 metadata SSRF, and an over-permissioned IAM policy.
Built and published Docker images, then demonstrated the supply-chain risk by containerizing a keylogger and capturing live keystrokes from it.
Compared SVM, Decision Tree, KNN, and Naive Bayes on IoT device classification, plus a manual PyTorch linear regression.
Decision Tree vs. Random Forest on 11,430 URLs, tuned via grid search. Random Forest reached 96% accuracy, 0.99 ROC-AUC.
Four models on 2.5M+ transactions at a 0.13% fraud rate. Random Forest hit 100% precision, 99.4% recall.
Five models classifying apps by permissions — the simplest one (Logistic Regression) won, missing only 2 of 42 malware apps.
Unsupervised clustering of 9,932 unlabeled PE files, validated with silhouette scores and a feature-redundancy stress test.
Acquired and hash-verified a forensic image of an 8GB drive, weighing logical vs. physical imaging and format trade-offs.
Full forensic workup of a suspect disk image: registry analysis, deleted files, malware indicators, browser history.
Recovered deleted evidence across four seized drives — including manually carving a PDF from raw hex signatures.
Incident-response critique of the 2021 ransomware attack — what failed, and the controls that would've prevented it.
Generated real MD5 collisions — including two executables that share a hash but run different code.
Forged a valid MAC against a naive keyed hash construction, then verified HMAC closes the hole.
Bypassed login and hijacked another user's account via UPDATE-statement injection, then fixed it with prepared statements.
Recreated the 2005 Samy MySpace worm — cookie theft, forged requests, and a self-propagating XSS worm.
Forged GET and POST requests to hijack a profile, then verified secret-token and SameSite-cookie defenses actually stop it.
Wrote sniffers, spoofers, and a traceroute from scratch with Scapy.
Built a firewall as a kernel module, then reproduced it with iptables — stateless, stateful, and rate-limited rules.
Poisoned ARP caches three ways, then intercepted and altered live Telnet traffic.
Decrypted DoH sessions with TLS session keys to compare plaintext vs. encrypted DNS visibility.
Reconstructed HTTP/1.1 conversations, then decrypted TLS 1.3 to compare HTTP/2 and HTTP/3 multiplexing.
I'm pursuing an M.Tech in Cybersecurity Engineering, building on a background in cybersecurity risk, compliance, and cloud security. My focus is the intersection of cloud infrastructure security, threat detection, and applying machine learning to security problems.
Certifications: Certified Ethical Hacker (CEH), AWS Certified Cloud Practitioner.