• Developed ATLAS - AI Security Testing Lab comprising of mainly two components: a Demo Lab and an automated GenAI Testing tool, based on OWASP Top 10 for LLM and ML and MITRE ATLAS frameworks. • ATLAS contains Model Extraction, Model Skewing, Adversarial Attack and Prompt Injection attack demo simulations improving security awareness and an automated Prompt Injection testing tool.• Involved in VAPT (Vulnerability Assessment and Penetration Testing) of applications in 2 client engagements, identifying and providing remediation of vulnerabilities found.• Participated in iterative Capture the Flag events.• Co-hosted a training session on 'Security for ML and ML for Security' for about 30 team members.

As the leader of the OSINT/Cyber Intel team, I was instrumental in testing, evaluating, classifying and implementing a comparative study of multiple open source tools and technologies from the Bellingcat Online Investigation toolkit and OSINT framework , thereby assimilating the knowledge of categorizing about 1500 OSINT tools into solving real-time live cybercrime cases which brought about considerable improvement in reaching conclusions. Along with this, I have come to have a deep rooted knowledge of emerging cyber threats, malwares and other vulnerabilities due to comprehensive and detailed analysis and cyber attribution reports prepared in my role. In the DeepWeb team, I was responsible for configuring and setting up tools such as OnionScan on virtual machines. OnionScan is useful for investigating the Dark Web by exploiting the most rudimentary and constantly prevailing threat of all-human error. My contributions in reconnaissance efforts related to credit card fraud, phishing and cyber attacks have helped the team identify and seek compensation for complainants of such cybercrime.

• Analyzed and identified an automation use-case harnessing the power of Splunk Querying in Splunk SIEM tool to drive data-driven insights and actions.• Played a pivotal role in automating threat notables using tools such as Splunk SIEM, Palo Alto Cortex XSOAR, Microsoft Defender 365.• My automation playbook of 'Detecting large outbound data packets' saves 56 hours per 90 days on average for the SOC Analysts triaging this particular threat alert. • Additionally learnt Kusto Querying in Advanced Threat hunting, Microsoft PowerApps, and Microsoft Power Automate.

•Developed a web scraper using Python and BeautifulSoup to gather data from surface web and dark web websites, by establishing connection with Tor browser using Firefox Geckodriver for secure access to dark web content•Cleaned and pre-processed the scraped data to prepare it for model training•Designed and implemented multi-class classification models to categorize dark web websites achieving an overall accuracy rate of 82% across categories such as Bitcoin, drugs, guns/weapons/rifles, and wiki.

Designed wireframes using Figma for intuitive user interfaces and developed an E-commerce website using ReactJs.

Developed an application to detect all devices present on a particular network along with the visible details of the device such as hostname, IPV4 address, MAC address, vendor name with graph plotting modules like NetworkX, plotly and Matplotlib to create a visual 'Network Map' resulting in a more efficient way to identify potential security threats.