Showing 6 of 6 projects
Blue Team
RAID-SecOps — AI-Powered Security Operations Platform
ML-driven alert triage with role-specific recommendations for SOC, Engineers & CISO
ProblemSIEM alerts reviewed inconsistently — SOC analysts, engineers and CISO/GRC had no unified system giving role-appropriate guidance on what to action and how
ActionBuilt role-aware FastAPI + PostgreSQL platform ingesting SIEM alerts; applied Random Forest for alert classification and Isolation Forest for anomaly detection — delivering tailored recommendations per role
OutcomeSOC analysts get triage steps, engineers get remediation actions, CISO/GRC get risk context — all from one platform, reducing manual decision overhead
🎯Add screenshot
Blue Team
Detection engineering — Microsoft Sentinel & Defender
Brute force reduced to zero; automated endpoint isolation in seconds
ProblemNo automated response for compromised endpoints; brute force attacks reaching systems through exposed Internet-facing ports
ActionAuthored custom KQL detection rules in Sentinel and MDE; implemented inbound NSG firewall rules to block unauthorised access vectors
Outcome100% reduction in brute force incidents; automated isolation triggered on compromise — response time cut from hours to seconds
🔍
t
Blue Team
tEnterprise vulnerability management programme
100% critical · 90% high · 76% medium vulnerabilities eliminated
ProblemServer infrastructure carried significant unresolved critical and high vulnerabilities across Windows and Linux, exposing the organisation to exploitation
ActionDeployed Tenable scans, DISA STIG compliance audits, and PowerShell-based automated remediations; prioritised using CVSS scoring
OutcomeFull elimination of critical vulnerabilities; measurable reduction across all severity tiers — significantly lowered organisational risk exposure
🕵️Add screenshot
Red Team
Threat hunting & IOC analysis — EDR investigations
Detected brute force, exfiltration & ransomware IOCs before escalation
ProblemHigh SIEM alert volume with poor signal quality; signature tools missing advanced threats in financial environment
ActionTriaged 30–40 alerts daily; built Splunk correlation rules; conducted proactive EDR behavioural threat hunts using MITRE ATT&CK framework
Outcome15% false positive reduction; ransomware and exfiltration activity detected prior to data loss — before any escalation
☁️Add screenshot
Cloud
Cloud security posture & access controls — Azure
40% compliant traffic improvement; 30% operational reliability uplift
ProblemPublic-facing financial systems had insufficient access controls and limited visibility into anomalous cloud traffic
ActionImplemented access control policies, Azure Monitor continuous monitoring, Syslog alerting, and ACL enforcement across network boundaries
Outcome40% improvement in compliant traffic; 30% reliability increase across IT systems — measurable uplift in security posture
🧠Add screenshot
Research
Novel Hybrid Deep Learning for Malware Classification — BSc Dissertation
94.8% accuracy on Microsoft BIG 2015 — 9 malware families classified
ProblemSignature-based AV tools fail against new and obfuscated malware variants — costing enterprises billions in breach costs annually
ActionDesigned MobileNet-V2 + CNN hybrid with primary input assortment strategy; converted PE binary files to grayscale images for deep feature extraction
Outcome94.8% accuracy, 95.7% precision, 94.7% sensitivity — outperforming single-model baselines at lower computational cost