I secure cloud systems, and the AI running inside them.
Cloud security and AI securityISO/IEC 27001 Lead AuditorU.S. Department of State ScholarFinal-year B.E. IT, Pune
Active Directory is where attackers aim first. I harden who gets in and what each account can reach.
SOC operations at EY, and Serverless SOAR, which turns Wazuh alerts into readable Slack triage.
Vulnerability assessment and penetration testing: finding the weakness before someone else does.
ISO/IEC 27001 Lead Auditor. Controls mapped across ISO 27001, PCI DSS and DPDP.
Deepfake and identity threat research, and Agent File Monitor, a tool I am building to log every file a coding agent touches.
I am a final-year B.E. Information Technology student in Pune. I spent a year in Virginia as a U.S. Department of State CCI Exchange Program scholar, then a year at EY working on SOC operations, penetration testing and Active Directory security.
My focus is how cloud systems get misconfigured and how AI models and agents get abused. I want to build the controls that catch both, and to audit them against the standards that matter.
Four projects, from alert triage to compliance mapping. The repositories are being rebuilt on GitHub.
Wazuh alerts travel through EventBridge and Lambda and land in Slack as readable triage.
Being built on AWS: it will check policy documents against ISO 27001 controls with retrieval and a foundation model, and cite the evidence behind each verdict.
Being built: it will log every file an AI coding agent touches and flag access outside its scope.
Crime forecasting that compares ARIMA, LSTM and a plain baseline on real incident data.
Two lines of work. One forecasts crime on real data, the other looks at how AI changes identity attacks.
Extends Natarajan et al. (2024) by testing a baseline, ARIMA and an LSTM side by side on Maharashtra data. The IEEE paper comes first. CrimeVision AI is then built on the same data, so the paper's results are the app's results.
Researched how synthetic voices and faces defeat identity checks, and which controls still hold. This sits next to my Active Directory work, since identity is the layer attackers cross first.
Ran SOC operations, carried out vulnerability assessment and penetration testing, and worked on Active Directory security. Researched AI and deepfake identity threats.
Cyber Security program funded by the U.S. Department of State. GPA 3.89 out of 4.00.
PCET NMIET, Savitribai Phule Pune University. Final year.
Issued by Mastermind Assurance. Lead auditor of information security management systems.
AIF-C01, Amazon Web Services.
Cyber Security program at Northern Virginia Community College.
Issued by The SecOps Group.
American Management Association, through NVCC. October 20, 2024.
Issued through Coursera.
Cloud security, AI security or research collaboration. Pune, Mumbai or remote.
© 2026 Dhruv Talegaonkar, Pune, India